Joseph K.T. Lee, Stuart S. Sagel, Robert J. Stanley, Jay P. Heiken - Computed Body Tomography With MRI Correlation-Lippin [PDF]

Zitiervorschau

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page i

Computed Body Tomography with MRI Correlation FOURTH EDITION

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page ii

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page iii

Computed Body Tomography with MRI Correlation FOURTH EDITION

EDITORS

JOSEPH K. T. LEE, MD E. H. Wood Distinguished Professor and Chair Department of Radiology University of North Carolina School of Medicine Chapel Hill, North Carolina

STUART S. SAGEL, MD Professor of Radiology Director, Chest Radiology Section Mallinckrodt Institute of Radiology Washington University School of Medicine St. Louis, Missouri

ROBERT J. STANLEY, MD, MSHA Editor-in-Chief American Journal of Roentgenology Professor and Chair Emeritus, Department of Radiology University of Alabama at Birmingham Birmingham, Alabama

JAY P. HEIKEN, MD Professor of Radiology Director, Abdominal Imaging Section Mallinckrodt Institute of Radiology Washington University School of Medicine St. Louis, Missouri

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page iv

Acquisitions Editor: Lisa McAllister Managing Editor: Kerry Barrett Project Manager: Fran Gunning Manufacturing Manager: Ben Rivera Marketing Manager: Angela Panetta Design Coordinator: Teresa Mallon Production Services: Nesbitt Graphics, Inc. Printer: Maple Press

© 2006 by LIPPINCOTT WILLIAMS & WILKINS 530 Walnut Street Philadelphia, PA 19106 USA LWW.com All rights reserved. This book is protected by copyright. No part of this book may be reproduced in any form or by any means, including photocopying, or utilized by any information storage and retrieval system without written permission from the copyright owner, except for brief quotations embodied in critical articles and reviews. Materials appearing in this book prepared by individuals as part of their official duties as U.S. government employees are not covered by the above-mentioned copyright. Printed in the USA Library of Congress Cataloging-in-Publication Data Computed body tomography with MRI correlation / editors, Joseph K.T. Lee, Stuart S. Sagel.— 4th ed. p. ; cm. Includes bibliographical references and index. ISBN 0-7817-4526-8 1. Tomography. 2. Magnetic resonance imaging. I. Lee, Joseph K. T. II. Sagel, Stuart S., 1940- . III. Title. [DNLM: 1. Tomography, X-Ray Computed. 2. Magnetic Resonance Imaging. WN 206 C7378 2005] RC78.7.T6C6416 2005 616.07’57—dc22 2005029421

Care has been taken to confirm the accuracy of the information presented and to describe generally accepted practices. However, the authors, editors, and publisher are not responsible for errors or omissions or for any consequences from application of the information in this book and make no warranty, expressed or implied, with respect to the currency, completeness, or accuracy of the contents of the publication. Application of this information in a particular situation remains the professional responsibility of the practitioner. The authors, editors, and publisher have exerted every effort to ensure that drug selection and dosage set forth in this text are in accordance with current recommendations and practice at the time of publication. However, in view of ongoing research, changes in government regulations, and the constant flow of information relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any change in indications and dosage and for added warnings and precautions. This is particularly important when the recommended agent is a new or infrequently employed drug. Some drugs and medical devices presented in this publication have Food and Drug Administration (FDA) clearance for limited use in restricted research settings. It is the responsibility of the health care provider to ascertain the FDA status of each drug or device planned for use in their clinical practice. 9 8 7 6 5 4 3 2 1

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page v

To our wives, Christina, Beverlee, Sally, and Fran To our children, Alexander, Betsy, and Catherine; Scott, Darryl, and Brett; Ann, Robert, Catherine, and Sara; and Lauren And to our grandchildren

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page vi

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page vii

Contents Contributing Authors ix Preface xi Acknowledgments xiii

1 BASIC PRINCIPLES OF COMPUTED TOMOGRAPHY PHYSICS AND TECHNICAL CONSIDERATIONS 1 Kyongtae T. Bae and Bruce R. Whiting 2 MAGNETIC RESONANCE IMAGING PRINCIPLES AND APPLICATIONS 29 Mark A. Brown and Richard C. Semelka 3 INTERVENTIONAL COMPUTED TOMOGRAPHY 95 Charles T. Burke, Matthew A. Mauro, and Paul L. Molina 4 NECK 145 Franz J. Wippold II 5 THORAX: TECHNIQUES AND NORMAL ANATOMY 225 Fernando R. Gutierrez, Santiago Rossi, and Sanjeev Bhalla

12 LIVER 829 Jay P. Heiken, Christine O. Menias, and Khaled Elsayes 13 THE BILIARY TRACT 931 Franklin N. Tessler and Mark E. Lockhart 14 SPLEEN 973 David M. Warshauer 15 THE PANCREAS 1007 Desiree E. Morgan and Robert J. Stanley 16 ABDOMINAL WALL AND PERITONEAL CAVITY 1101 Jay P. Heiken, Christine O. Menias, and Khaled Elsayes 17 RETROPERITONEUM 1155 David M. Warshauer, Joseph K. T. Lee, and Harish Patel 18 THE KIDNEY AND URETER 1233 Mark E. Lockhart, J. Kevin Smith, and Philip J. Kenney 19 THE ADRENAL GLANDS 1311 Suzan M. Goldman and Philip J. Kenney

6 MEDIASTINUM 311 Alvaro Huete-Garin and Stuart S. Sagel

20 PELVIS 1375 Julia R. Fielding

7 LUNG 421 Stuart S. Sagel

21 COMPUTED TOMOGRAPHY OF THORACOABDOMINAL TRAUMA 1417 Paul L. Molina, Michele T. Quinn, Edward W. Bouchard, and Joseph K. T. Lee

8 PLEURA, CHEST WALL, AND DIAPHRAGM 569 David S. Gierada and Richard M. Slone 9 HEART AND PERICARDIUM 667 Pamela K. Woodard, Sanjeev Bhalla, Cylen Javidan-Nejad, and Paul D. Stein 10 NORMAL ABDOMINAL AND PELVIC ANATOMY 707 Dennis M. Balfe, Brett Gratz, and Christine Peterson 11 GASTROINTESTINAL TRACT 771 Cheri L. Canon

22 MUSCULOSKELETAL SYSTEM 1481 Robert Lopez-Ben, Daniel S. Moore, and D. Dean Thornton 23 THE SPINE 1661 Zoran Rumboldt, Mauricio Castillo, and J. Keith Smith 24 PEDIATRIC APPLICATIONS 1727 Marilyn J. Siegel Index 1793

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page viii

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page ix

Contributing Authors

Associate Professor of Radiology, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri

Julia R. Fielding, MD

Professor of Radiology, Department of Diagnostic Radiology, Washington University School of Medicine, St. Louis, Missouri

David S. Gierada, MD

Assistant Professor of Radiology, Co-Chief, CT and Emergency Radiology, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri

Suzan Menasce Goldman, MD, PhD

Kyongtae T. Bae, MD, PhD

Dennis M. Balfe, MD

Sanjeev Bhalla, MD

Radiology Resident, University of North Carolina School of Medicine, Chapel Hill, North Carolina

Edward W. Bouchard, MD

Senior Technical Instructor, Siemens Training and Development Center, Cary, North Carolina

Mark A. Brown, PhD

Associate Professor and Director of Abdominal Imaging, Department of Radiology, University of North Carolina School of Medicine, Chapel Hill, North Carolina Associate Professor of Radiology, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri

Affiliated Professor, Imaging Diagnosis Department, UNIFESP/EPM, São Paulo, Brazil

Instructor in Radiology, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri

Brett Gratz, MD

Professor of Radiology, Cardiothoracic Imaging Section, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri

Fernando R. Gutierrez, MD

Assistant Professor of Radiology, University of North Carolina School of Medicine, Chapel Hill, North Carolina

Jay P. Heiken, MD

Associate Professor, Vice Chair for Education, Department of Radiology, University of Alabama at Birmingham; Chief, Gastrointestinal Radiology, Department of Radiology, UAB Health System, Birmingham, Alabama

Alvaro L. Huete-Garin, MD

Charles T. Burke, MD

Cheri L. Canon, MD

Professor and Director of Neuroradiology, Department of Radiology, University of North Carolina School of Medicine, Chapel Hill, North Carolina

Mauricio Castillo, MD

Khaled M. Elsayes, MD

Institute, Giza, Egypt

Staff Radiologist, Theodore Bilhars

Professor of Radiology, Department of Radiology, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri Assistant Professor of Radiology, Catholic University, Santiago, Chile

Assistant Professor of Cardiothoracic Imaging, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri

Cylen Javidan-Nejad, MD

Director of Outpatient Radiology and Chief, GU Section, Professor, Abdominal Imaging Section, Department of Radiology, University of Alabama at Birmingham, Birmingham, Alabama

Philip J. Kenney, MD

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page x

x

Contributing Authors E. H. Wood Distinguished Professor and Chairman, Department of Radiology, University of North Carolina School of Medicine, Chapel Hill, North Carolina

Joseph K. T. Lee, MD

Director, Abdominal Imaging Fellowship, Assistant Professor, Abdominal Imaging Section, Department of Radiology, University of Alabama at Birmingham, Birmingham, Alabama

Mark E. Lockhart, MD, MPH

Associate Professor of Radiology, University of Alabama Medical School, Birmingham, Alabama

Robert Lopez-Ben, MD

Professor and Vice Chair of Clinical Affairs, Department of Radiology, University of North Carolina School of Medicine, Chapel Hill, North Carolina

Matthew A. Mauro, MD

Assistant Professor, Department of Radiology, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri

Christine O. Menias, MD

Professor of Radiology and Vice Chairman of Education, Department of Radiology, University of North Carolina School of Medicine, Chapel Hill, North Carolina

Paul Lee Molina, MD

Assistant Professor, Department of Radiology, University of Texas Southwestern Medical School, Dallas, Texas

Daniel S. Moore, MD

Associate Professor and Medical Director—MRI, Department of Radiology, University of Alabama at Birmingham, Birmingham, Alabama

Desiree E. Morgan, MD

Clinical Instructor, Department of Radiology, University of North Carolina School of Medicine, Chapel Hill, North Carolina

Harish Patel, MD

Christine M. Peterson, MD Clinical Fellow, Department of Radiology, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri Radiology Resident, University of North Carolina School of Medicine, Chapel Hill, North Carolina

Michele T. Quinn, MD

Centro de Diagnostico, Hospital de Clínicas José de San Martín, Buenos Aires, Argentina

Santiago Enrique Rossi, MD

Associate Professor of Radiology, Medical University of South Carolina, Charleston, South Carolina

Zoran Rumboldt, MD

Professor of Radiology and Director, Chest Radiology Section, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri

Stuart S. Sagel, MD

Professor and Vice Chair of Research, Department of Radiology, University of North Carolina School of Medicine, Chapel Hill, North Carolina

Richard C. Semelka, MD

Professor of Radiology and Pediatrics, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri

Marilyn Joy Siegel, MD

Virtual Radiologic Professionals, PLLC, Virtual Radiologic Consultants, Minneapolis, Minnesota

Richard M. Slone, MD, FCCP

Vice Chair for Veterans Affairs, Associate Professor, Abdominal Imaging Section, Department of Radiology, University of Alabama at Birmingham, Birmingham, Alabama

J. Kevin Smith, MD, PhD

Associate Professor of Radiology, University of North Carolina School of Medicine, Chapel Hill, North Carolina

J. Keith Smith, MD, PhD

Professor and Chair Emeritus, Department of Radiology, University of Alabama at Birmingham, Birmingham, Alabama

Robert J. Stanley, MD, MSHA

Paul D. Stein, MD

St. Joseph Mercy Hospital, Pontiac,

Michigan Professor of Radiology, Department of Radiology, University of Alabama at Birmingham, Birmingham, Alabama

Franklin N. Tessler, MD, CM

Clinical Assistant Professor, Department of Radiology, University of Alabama Medical School, Birmingham, Alabama

D. Dean Thornton, MD

Professor of Radiology, University of North Carolina School of Medicine, Chapel Hill, North Carolina

David M. Warshauer, MD

Research Assistant, Professor of Radiology, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri

Bruce R. Whiting, PhD

Professor of Radiology, Chief of Neuroradiology, Mallinckrodt Institute of Radiology, Washington University Medical Center, St. Louis, Missouri; Adjunct Professor of Radiology and Nuclear Medicine, F. Edward Hébert School of Medicine, Uniformed Services University of the Health Sciences, Bethesda, Maryland

Franz J. Wippold II, MD, FACR

Associate Professor, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri

Pamela K. Woodard, MD

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page xi

Preface Since the publication of the third edition of our textbook Computed Body Tomography with MRI Correlation in 1998, major technologic advances have been made in both computed tomography (CT) and magnetic resonance imaging (MRI). The evolution from a single-detector-row helical (spiral) CT to multidetector-row CT (MDCT) has provided the unique opportunity to perform isotropic volumetric imaging and allowed new clinical indications. CT angiography now is routinely used for the detection of pulmonary emboli, for assessment of the aorta and its branches, for preoperative planning for resection of selected thoracic and abdominal tumors, and prior to donor nephrectomy. The 64 MDCT scanner now has replaced the electron beam scanner for assessing the coronary arteries as well as the cardiac anatomy and function. CT has become the procedure of choice for evaluating patients with acute abdominal pain and multiorgan trauma. Although controversial, largely because of cost–benefit and radiation-dose issues, CT also has been used to screen asymptomatic individuals in some centers. The development of PET-CT combines the metabolic information provided by PET with superb anatomic resolution provided by CT. PETCT has now become an integral part of oncologic imaging. During the same period of time, innovations and refinement in MR hardware and software technology have continued. Faster pulse sequences, improved coil design, and the development of parallel imaging all have contributed to the increased utilization of MR as a diagnostic tool. MRI is clearly the procedure of choice for evaluating many diseases of the central nervous system and the musculoskeletal system. Although MRI is well suited for assessing the cardiovascular system and has the advantage of not using ionizing radiation, the clear superiority of MRI over CT for imaging the cardiovascular system that was so evident several years ago is less apparent now because of the development of 64 MDCT scanners. However, MRI has

been well established as a complementary imaging study in the abdomen and pelvis. The role of MRI in thoracic imaging is still limited. This edition has been prepared to present a comprehensive text on the application of CT to the extracranial organs of the body. The role of MRI in these areas is also fully discussed, wherever applicable. The book is intended primarily for the radiologist to use in either clinical practice or training. Other physicians, such as the internist, pediatrician, and surgeon, also can derive state-of-the-art information about the relative value and indications for CT and MRI of the body. As in the first three editions, both normal and abnormal CT and MRI findings are described and illustrated. Instruction is provided to optimize the performance, analysis, and interpretation of CT and MR images. Information is provided on how to avoid technical and interpretative errors commonly encountered in CT and MRI examinations based on our collective experience. The task of deciding which diagnostic test is most appropriate for a given clinical problem has remained a challenge in our practice. A thorough understanding of clinical issues, as well as the advantages and limitations of each imaging technique, is essential for determining the best imaging approach for establishing a specific diagnosis in a given situation. Our recommended uses of CT and MRI have been developed through the efforts of radiology colleagues at our three medical centers. We are fully aware that equally valid alternative imaging approaches to certain clinical problems exist. Furthermore, increasing knowledge, continued technologic improvement, and differences in available equipment and expertise will influence the selection of a particular imaging method at a given institution. J.K.T.L. S.S.S. R.J.S. J.P.H.

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page xii

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page xiii

Acknowledgments Providing recognition to everyone involved in the production of this edition is extremely difficult because of the large number of individuals from our three institutions who aided immeasurably in forming the final product. We graciously thank the various contributors who kindly provided chapters in their areas of expertise to bring depth and completeness to the book. A special note of gratitude goes to our secretaries, Sue Day, Angela Lyght, Jama Rendell, Pam Schaub, and Trish Thurman, who spent endless hours typing manuscripts, checking references, and labeling images. Maurice Noble at the University of North Carolina Department of Radiol-

ogy Photography Laboratory was extremely helpful in preparing the illustrative material. Our thanks go to our residents, fellows, and the many radiologic technologists who performed and monitored the CT and MRI studies. Their dedication is reflected in the high quality of the images used throughout this book. We also would like to express our appreciation to Lippincott Williams and Wilkins for their professionalism in handling this project. Most particularly, we would like to thank Kerry Barrett and Lisa McAllister for their tireless dedication and advice during each stage in the production of this book.

5063_Lee_FMppi-xiv 10/20/05 12:44 PM Page xiv

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 1

Basic Principles of Computed Tomography Physics and Technical Considerations Kyongtae T. Bae

1

Bruce R. Whiting

INTRODUCTION Slightly more than three decades old, computed tomography (CT) continues to advance rapidly in both imaging performance and widening clinical applications. An appreciation of the potential of CT and its limitations can be obtained with an understanding of basic principles of CT operations. This chapter provides background and insight into the technical issues surrounding the application of CT, including the image formation process, various parameters affecting clinical usage, metrics to describe performance, the display of image information, and radiation dose.

Imaging with X-Rays X-ray imaging was the first diagnostic imaging technology, invented immediately after the discovery of x-rays by Roentgen in 1895. X-rays are a form of electromagnetic energy that propagate through space and are absorbed or scattered by interactions with atoms. The attenuation of beam energy on passage through physical objects provides a noninvasive means to gather information about the amount and type of material present inside the object.

In radiography, x-rays illuminate an object, resulting in a two-dimensional (2D) image that is the “shadow” of three-dimensional (3D) structures present in the beam. The projection causes a superposition of internal structures, leading to indeterminacy in the exact relationships, shapes, and relative positions of objects. Because of this indeterminacy, radiologists require extensive training and experience to interpret 3D structures from the 2D image data. Furthermore, projection radiographs have very limited ability to differentiate low-contrast differences in tissues. Computed tomography (CT) was created in the early 1970s to overcome many of these limitations (13). By acquiring multiple x-ray views of an object and performing mathematical operations on digital data, a full 2D section of the object can be reconstructed with exquisite detail of the anatomy present (Fig. 1-1). During the years since its invention, CT technology has undergone continual improvement in performance through refinements in components and innovation in scanning techniques (19). As a result, scan times have dramatically improved, and volume coverage and resolution detail have increased.

1

5063_Lee_Ch01pp0001-0028 2/17/06 9:36 AM Page 2

2

Chapter 1

A

B Figure 1-1 A: X-ray and B: computed tomography of head with cochlear implant. Note the higher contrast of fine structures in the computed tomography slice, whereas superposition of structures in the x-ray confounds the three dimensional location.

As a curious consequence of this progress, the very large volume of image data acquired with current scanning techniques poses another challenge for interpretation: how to display very large amounts of information for the interpretation process. The magnitude and complexity of true volume imaging requires new rendering techniques to enable productive exploitation of the vast amount of information.

1 second, with reconstruction computations requiring several seconds per slice. Nevertheless, the time required to scan a patient volume of interest often was longer than a single breath-hold, and scan range was limited by x-ray tube heat load to 10 to 30 cm. By translating the patient table continuously through the rotating gantry, termed helical or spiral scanning, volume coverage and scan speed were further increased, with fundamental rate limitations being x-ray

Brief History of Computed Tomography Evolution of CT Performance 108 107 Acquired Pixels per Second

Since its introduction in the mid-1970s, CT scanner technology has undergone a continual improvement in performance, including increases in acquisition speed, amount of information in individual slices, and volume of coverage. A graph (Fig. 1-2) of these parameters versus time looks similar to Moore’s Law for computer priceperformance, which observes that computer metrics (clock speed, cost of random access memory or magnetic storage, etc.) double every 18 months. In the case of CT technology, the doubling period is approximately 32 months, still an impressive rate. For example, scan time per slice has decreased from 300 seconds in 1972 to 0.005 seconds in 2005. Factors contributing to this remarkable advance include improvements in electronics hardware and development of innovative mechanical scanning configurations. Historically, the early scanner configurations were characterized as successive generations of scanner geometry (Fig. 1-3). By 1990 rotating fan beam systems, utilizing slip-ring technology to allow continuous rotation of x-ray tube and detector, had reduced acquisition time to about

106 105 104 103 102 1970

1975

1980

1985 1990 Date

1995

2000

2005

Figure 1-2 Evolution of computed tomography scanner performance: plot of acquisition performance versus time, for computed tomography scanners. The slope implies a doubling of performance approximately every 2 years. (Data from Siemens Medical Systems, www.medical.siemens.com, “CT History and Technology.”)

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 3

Basic Principles of Computed Tomography Physics and Technical Considerations 1st generation (1970)

3

2nd generation (1972) 1. translation

translation start

end

start end

Pencil beam: translation/rotation

Partial fan beam: translation/rotation

3rd generation (1976)

4th generation (1978)

rotating detector arc

Fan beam: continuous rotation

stationary detector ring Fan beam: continuous rotation

tube output and mechanical rotation rate. Image reconstruction techniques were developed to interpolate 2D planes from the 3D datasets that were acquired in helical mode. In the late 1990s, the obstacles encountered by early helical scanners were overcome by multidetector row technology, using multiple sets of detector rows to utilize more of the x-ray tube output and acquire measurements at multiple section levels in parallel. Reconstruction under these conditions is inherently 3D, so more complex algorithms must be used. Benefiting from substantial improvements in computing power, the rapid increases in CT performance appear to be sustainable into the new century, with development of flat panel detectors, faster electronics, and cone-beam geometry reconstruction algorithms. To understand best how to utilize CT technology clinically and appreciate new product capabilities, knowledge

Figure 1-3 Definition of the different generations of scan geometries.

of fundamental CT imaging principles is necessary. The basic principles of CT involve physical mechanisms that are shared with x-ray imaging, plus mathematical techniques that exceed the human visual perception of 2D images. A common technical description can be used to describe both the image formation process and the image visualization task. These will now be examined in detail.

COMPUTED TOMOGRAPHY ACQUISITION SYSTEM COMPONENTS Generation of X-Rays For medical imaging, x-rays are generated by an x-ray tube. In this device, a metal filament is heated (much like a light bulb) until energetic electrons escape from the cathode

5063_Lee_Ch01pp0001-0028 2/17/06 9:37 AM Page 4

4

Chapter 1 0.08 0.07

Probability

0.06 0.05 0.04 0.03 0.02 0.01 0

100

50

150

Energy (keV)

Figure 1-4 Typical x-ray spectrum for tungsten target with 120 kVp. Low-energy photons, which do not pass through the patient to contribute to final image, have been filtered out.

surface into a vacuum. These electrons are then accelerated by an electric field, acquiring kinetic energy while being attracted to a positive anode target. The total amount of energy acquired by the electron in the accelerating electric field is equal to the product of the potential (peak kilovoltage, kVp) times the unit of electrical charge, possessing units of electron volts (kilo electron volts, keV). The amount of charge generated by the x-ray tube per unit time has units of electrical current (milliamperes, mA), and the product of voltage and current is the amount of power (watts) delivered by the tube. Electrostatic and/or magnetic fields are used to focus the electron beam into a small area of the anode target. Typically, this focal spot has dimensions of about 1 mm. When the electrons collide with the target, most of their energy is dissipated into heat but a small fraction (
1%) is converted into several forms of electromagnetic radiation. A typical spectrum of the distribution of energy emitted by the x-ray tube is shown in Figure 1-4. Characteristically, there is a linearly decreasing portion caused by bremsstrahlung, the deceleration of the electrons in the target. According to Maxwell’s equations, any charge undergoing acceleration will radiate electromagnetic energy. As beam electrons pass through target atoms, they interact and are accelerated. The maximum amount of energy that can be transferred is equal to e  kVp, and lesser amounts of energy appear randomly depending on the details of electron collisions. The sharp peaks in the spectrum occur when the beam electrons deposit energy by exciting atomic electrons in the target. Electron shell transfers arise in atoms, with characteristic radiation at well-defined (K-edge) energy peaks. The spectrum generated in an x-ray tube contains many low energy photons. The power in the beam associated with a particular energy range is fairly constant, because the number of quanta decreases linearly as a function of

energy, while the energy of an individual quantum increases linearly. Because the lowest energy quanta are effectively attenuated in the patient, they contribute very little to the measured signal while exposing the patient to radiation dose. Therefore, the beam is filtered by placing material around the x-ray tube to reduce much of the low energy quanta while passing high energy quanta, leading to an optimal image quality/dose tradeoff. The x-rays from the target are spread over a wide solid angle (essentially a hemisphere). To minimize radiation dose and generation of background scatter, the x-ray beam is collimated by an aperture into a thin fan beam. For CT scanners, the beam is typically a few millimeters thick in the patient, subtending a fan of about 45 degrees. Additionally, because human anatomy typically has a round cross-section that is thicker in the middle than in the periphery, more x-ray flux reaches detectors on the edges than at the center. This means that patients receive more dose than is necessary on the periphery of their anatomy. To compensate for this effect, a bowtie-shape filter is placed in the beam, which is tapered such that its center is thinner than its edges, to equalize the flux reaching the detectors and minimize patient dose. The inefficiency in conversion of electron current into x-rays has been a significant practical limitation in the operation of x-ray imaging equipment. The tube is quickly heated to high temperatures, which must be limited to avoid damage. Anode targets have been designed to rotate on bearings, spreading out the area that is heated by the beam. Heat sinks are used to remove heat from the system by convection or water-assisted cooling. In typical clinical operation, an x-ray tube delivers on the order of 2  1011 x-rays per second to the patient, providing a high signal-to-noise ratio for measurements.

Detection of X-Rays Detection of x-rays is accomplished by the use of special materials that convert the high energies (tens of keV) of the x-ray quantum into lower energy forms, such as optical photons or electron-hole pairs, which have energies of a few electron volts. In this down-conversion, many secondary quanta are generated, typically thousands per primary quanta. The detector materials, such as phosphors, scintillating ceramics, or pressurized xenon gas, ultimately produce an electrical current or voltage. Electronic amplifiers condition this signal, and an analog-to-digital converter converts it into a digital number. The range of signals produced in tomography is large, varying from a scan of air (no attenuation, or 100% transmission) to that of a large patient with metal implants (possible attenuation of 0.0006%), a factor of almost 105. Furthermore, even at the lowest signal levels, the analog-to-digital converter must be able to detect modulations of a few percent. Thus the overall range approaches a factor of one million, specifying the equivalent of a 20-bit analog-to-digital converter.

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 5

Basic Principles of Computed Tomography Physics and Technical Considerations

Gantry Electromechanics To obtain required measurements at different angles, all the electrical components must be rotated around the patient. In modern scanners, this puts tremendous requirements on mechanical precision and stability. The gantry can weigh 400 to 1,000 kg, span a diameter of 1.5 m, and rotate 3 revolutions per second. While rotating, it may not wobble more that 0.05 mm. Originally, the gantry was connected by cables to the outside environment and had to change rotation direction at the end of each revolution. A major breakthrough in scanning operation occurred with the invention of slip-ring technology, which used brush contacts to provide continuous electrical power and electronic communication, allowing continuous rotation.

Helical/Spiral Scanning One of the primary goals of CT manufacturers has been to provide faster scan times and larger scan coverage. With the advent of slip-ring technology and continuous gantry rotation, the main limitation to scanning speed was the stepping of the patient bed to position sequential slices. In the late 1980s continuous motion of the patient table was introduced, which allowed faster scan times but required different data handling for image reconstruction (Fig. 1-5). Previously the theory of CT reconstruction was based on having a complete set of gantry measurements for each slice reconstructed. However, in helical scans the gantry is at continuously different table positions throughout each rotation. A good mathematical approximation for each gantry position is to interpolate a reconstruction plane from corresponding neighboring gantry positions. This approach provided adequate image quality, and in fact had the added benefit that slices could be reconstructed retrospectively for arbitrary table positions, instead of being limited to fixed table increments. Furthermore, analysis revealed that on average the spatial resolution was better

5

with helical scans rather than sequential scans. A drawback was that the interpolation process could create stair-step artifacts on the boundaries of extended high-contrast objects.

Detector Configuration By the mid-1990s, helical scans had become limited in speed because of the mechanical forces associated with subsecond gantry rotation times and the output requirements of x-ray tubes to supply enough flux for adequate signal to noise ratio. The next improvement in performance resulted from acquiring measurements at multiple body levels in parallel, using more than one row of detectors at the same time. This advance allowed an increase in speed of volume acquisition proportional to the number of rows of detectors. In this approach, the x-ray tube produces a broad beam of x-rays, rather than one that is collimated to a narrow slice; by widening the collimation to illuminate multiple rows of detectors, more measurements are acquired from the same tube output. Initially, two- or four-row multidetector row CT (MDCT) scanners were introduced, but the number of detector rows has grown steadily, with 64-detector row devices now enabling very large volume coverage. Because of the increased longitudinal width of the x-ray beam with MDCT, image data measurements no longer correspond to rays orthogonal to the scan axis; thus new reconstruction algorithms are required to maintain image quality and prevent distortions. In single-detector row CT (SDCT), each individual detector row functions as a single unit and provides projection data for a single section per rotation. In SDCT, different section widths are obtained by means of adjusting prepatient collimation of the x-ray beam (Fig. 1-6). In MDCT, the detectors are further divided along the z-axis, allowing simultaneous acquisition of multiple sections per rotation. Thus MDCT provides larger and faster z-axis coverage per rotation with thinner section widths.

Path of continuously rotating x-ray tube and detector

Start spiral scan

Direction of patient transport 0 Start 0

z, mm t, s

Figure 1-5 Helical or spiral scanning involves translating the patient longitudinally through the rotating gantry.

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 6

6

Chapter 1 Focus

X-ray focus

Collimator

Scan-FOV

Scanfield 16x1.25

z-axis

z-axis

Fixed Array Detector

Detector

A

X-ray focus 5

2.5

1.5

1

1

1.5

X-ray focus 2.5

5

Scanfield

Scanfield 4x1.5

z-axis

16x0.75

4x1.5

Adaptive Array Detector

C

B

16 rows, 4 slices

Slice-Width

8 rows, 4 slices

24 rows, 16 slices

z-axis

Adaptive Array Detector D

Figure 1-6 Multidetector computed tomography configurations. A: Illustration shows prepatient collimation of the x-ray beam to obtain different collimated section widths with a single-detector row computed tomography detector. FOV  field of view. Illustrations show examples of (B) fixedarray and (C, D) adaptive-array detectors used in commercially available four-section and 16-section computed tomography systems.

When four-channel MDCT scanners were introduced in the late 1990s, three different detector configurations were used by the CT manufacturers: (A) 16 detector rows with a uniform thickness, termed uniform array (General Electric); (B) eight detector rows of variable thicknesses, thinner rows centrally and wider rows peripherally, termed adaptive array [Siemens and Philips]; and (C) 34 detector rows with two fixed thicknesses, four thinner rows centrally and 30 thicker rows peripherally, termed hybrid array (Toshiba). Note that four-channel MDCT systems contain detectors that are divided into eight to 34 rows along the z-axis. Nevertheless, the number of sections acquired at each rotation is restricted to four because these systems contain only four data channels. When a scan with a narrow collimation is desired, four individual central detector rows are used for the data measurement, with a narrowly collimated x-ray beam directed over these central detector rows (e.g., 4  1 mm). To generate scans with larger section widths, a broadly collimated x-ray beam is used, and outputs from two or more adjacent detector rows are electronically combined into a single thicker detector row for each of the four data channels. For example, two 1-mm detector rows can be grouped to function as a single detector row for 2-mm collimation (4  2 mm), three 1-mm detector rows for 3-mm collimation (4  3 mm), and so on.

For 16-channel MDCT, all of the CT manufacturers adopted a hybrid array design, in which the thickness of the detector rows is slightly less than 1 mm for the central rowsand slightly more than 1 mm for the peripheral rows. However, the length of the z-axis coverage and the number of detector rows varies widely among the CT manufacturers. For 64-channel MDCT, the CT manufacturers have again used a common detector row design, this time a uniform array in which all the detector rows have a uniform thickness. However, as in 16-channel MDCT, the total number of detector rows and the z-axis coverage are highly variable among the CT manufacturers.

COMPUTED TOMOGRAPHY IMAGE FORMATION X-Ray Signals X-ray imaging consists of the generation of x-rays, transmission of those x-rays through material objects, and the detection of the beam energy that exits the object. The attenuation of x-rays within an object is governed by interactions on the atomic scale, in which each molecule in the object has some cross section for interacting with each x-ray. Because of this interaction, the x-ray flux decreases on

5063_Lee_Ch01pp0001-0028 2/17/06 9:37 AM Page 7

Basic Principles of Computed Tomography Physics and Technical Considerations

average by a certain percentage for each unit distance traveled through the object. Thus, if a 60 keV x-ray travels through 1 mm of water, on average it will survive 97.4% of the time. For 2 mm of water, the survival probabilities multiply for a 95% rate. The transmission probability is thus an exponentially decreasing function of the total amount and type of material present, represented by Lambert-Beer equation: S  I expaa
iti b

(1)

i

where S is the number of surviving signal quanta, I is the number of incident quanta, the subscript i indicates different materials that are compose the sample,
i is the linear attenuation coefficient for each material and ti is the amount (thickness) of that material present. In projection x-ray imaging, the image consists of the relative changes in the signal S across a viewing area. For a 70-kg person, with an abdomen roughly equivalent to 20-cm thickness of water, the survival probability for a single quantum would be about 2%. The presence of an additional 2 mm of abnormal structure would change this survival probability to 1.98% (only a 1% difference). Given this small change in the midst of many overlapping body structures, it is clear that projection radiography is limited in its ability to demonstrate anatomic details. In CT imaging, measurements of S are made from multiple projections, and from these measurements
i is computed for direct display. This technique results in much higher relative contrast between adjacent structures. For example, a 2-mm calcified nodule may have a 200% difference in attenuation coefficient compared with surrounding tissue, and hence be much more conspicuous than on a projection radiograph (see Fig. 1-1). For the viewing of images, projection x-rays are presented as a brightness that is proportional to the changes of the transmitted signal S in Eq. 1. In CT, the image attenuation map is presented in units that are relative to the attenuation coefficient of water, expressed as Hounsfield units (HU).
i 
water HUi
1,000 (2)
water

7

Image Reconstruction From Two-Dimensional Projection Data The basics of CT image generation can be illustrated by the reconstruction of a 2D image section from projection measurements. An x-ray source and a set of detectors rotate around the patient, making measurements of the transmission of x-rays through the body. Each measured value is the result of all the attenuating portions in the patient along a line from the x-ray source to the detector making the measurement. Hence, a uniform circular disk will have highest attenuation in its center, with a circular profile. The collection of line measurements from different view angles during one revolution of the gantry provides raw projection data prior to reconstructing images. The raw projection data result in a sinogram (Fig. 1-7). The sinogram can be displayed as an image, with the y-axis (rows) representing the measurements of each detector and the x-axis (columns) representing detector measurements at one gantry position. The sinogram image has an intriguing pattern, but is difficult to interpret because of the overlapping shapes. Thus, a method is needed to derive and compute the original image attenuation. One method, albeit impractical, for determining the source image involves treating the sinogram and image as a linear algebra problem. Each measurement is an equation summing all the image pixels along a ray to the detector; the set of all equations can then be solved for the image pixel unknowns. The size of this problem is dauntingly large because there are 512  512 (i.e., more than one quarter million) variables involved with 768  1,400 (i.e., more than one million) measurements, requiring matrix operations that overwhelm even modern computers. Other mathematical methods, such as iterative techniques or maximum likelihood optimization, can be used to solve for images, but they also are too computationally intensive for routine clinical usage. The mathematical process that made CT reconstruction practical is called filtered back projection. It can be shown theoretically (18) that if the projection measurements have certain properties (they all lie in one plane, they con-

A

B Figure 1-7 An abdominal slice and its sinogram.

5063_Lee_Ch01pp0001-0028 2/17/06 9:37 AM Page 8

8

Chapter 1

sist of equally spaced gantry steps covering at least one half revolution, and the detectors are equidistant and cover the whole object to be reconstructed), then the attenuation (image) at any point within the scanner field of view can be calculated by summing a certain weighted combination of the measurements. This weighted summation process is called a kernel (see the section titled Reconstruction Kernel later in this chapter for detail). The measurement of the detector directly intercepting the pixel is added and measurements from neighboring detectors are subtracted. Different kernels can be designed to provide sharp, crisp images or to smooth out noise, depending on the clinical application. This process, which was universally adopted by CT manufacturers in the early years of CT, can be performed very efficiently by computers or special hardware modules, either directly or with Fast Fourier Transform techniques.

Image Reconstruction from Three-Dimensional Projection Data The filtered back projection process requires that the image data be confined to a single plane. With helical CT, 3D volumes rather than single sections of data are acquired, necessitating the development of new reconstruction algorithms.

Single-Detector Row Spiral Computed Tomography: Linear Interpolation In spiral scanning, the patient table moves continuously, so at any given longitudinal or z-location there are only a few (or no) exactly corresponding gantry measurements that are aligned in the same plane for 2D filtered back projection. The higher the pitch (i.e., the faster the CT table travels relative to the detector collimation), the more the gantry measurements separate and deviate from the plane. To provide a complete set of measurements for filtered back projection, missing gantry measurements are estimated by taking the average of the closest (in the z-axis) measurements that are collected. Two versions of this method are employed. The first is called 360LI and takes averages of measurements separated by one rotation. In this approach, to generate projection data for a target image plane, two gantry measurements on either side of the image plane that are positioned closest to the image plane and are 360 degrees apart (i.e., are measured in subsequent rotations) are linearly interpolated for each projection angle. The 360LI technique has the disadvantage that the travel in one revolution may be large, and if structures change significantly over this distance blurring or partial volume averaging will result. The second method, called 180LI, takes advantage of the symmetry between x-ray source and detector across the gantry, i.e., the measured ray is nominally the same when the source-detector positions are one half rotation (180 degrees) apart. The 180LI technique makes use of the fact that for each measurement ray, an interpolation

partner is already available after approximately one half a rotation, when the x-ray tube and detector have switched positions. This virtual, geometrically derived ray is called a complementary ray. The 180LI technique involves smaller z-distances and hence suffers less blurring. (The same trick can be used in cardiac imaging, to shorten the time window for an image snapshot and minimize temporal blur.)

Multidetector Row Spiral Computed Tomography: Z-Interpolation or Z-Filtering The first multidetector row scanners had two or four detector rows, and the data measurements could be treated as a simple parallel stack of independent detector rows. In this case, the 360LI and 180LI used in SDCT spiral reconstruction approaches can be directly extended to spiral MDCT. One could then create planes of measurements by linear interpolation (either 360LI or 180LI) from the closest row measurements to the target plane, a technique known as advanced single-slice rebinning (16). The interpolation calculation can be performed very rapidly and is essentially similar to single-row scanning. In the 360LI interpolation approach, the interpolation can be performed using rays measured at the same projection angle by different detector rows or in consecutive rotations of the scanner 360 degrees apart. In the 180LI reconstruction approach, both direct and complementary rays can be used for spiral interpolation. CT scanner manufacturers proposed different mathematical approaches for weighting and interpolating neighboring rays for the target image plane, such as z-interpolation or z-filtering (17,32,34).

Broad Beam Multidetector or Flat-Panel Computed Tomography: Cone Beam Reconstruction With increases in the number of detector rows beyond four, it becomes necessary to account for the cone-beam angle between detector rows (8). Some manufacturers use variations and extensions of nutating-section algorithms for image reconstruction (4,16,21,31). These algorithms split the 3D reconstruction task into a series of conventional 2D reconstructions on tilted intermediate image planes, thereby benefiting from established and very fast 2D reconstruction techniques. Examples are adaptive multiplanar reconstruction (Siemens) (7) and the weighted hyperplane reconstruction (GE Medical Systems) (12) techniques. Other manufacturers (Toshiba, Philips) have extended to multisection scanning the Feldkamp algorithm (6,9), an approximate 3D convolution back-projection reconstruction that was originally introduced for sequential scanning. With this approach, accounting for their conebeam geometry, the measurement rays are back projected into a 3D volume along the lines of measurement. Three-dimensional back projection is, however, computationally demanding and requires dedicated hardware to achieve acceptable image-reconstruction times. The development of methods to account for the cone-beam geometry of the measurement x-rays currently is an active area of research.

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 9

Basic Principles of Computed Tomography Physics and Technical Considerations

IMAGING METRICS Although image quality is the ultimate measure of an imaging system, it is difficult to define and quantify image quality. In clinical settings, image quality is frequently determined qualitatively and subjectively. Communication theory specifies the fundamental parameters of information transfer as signal, resolution, distortion, and noise to characterize system performance. Several quantitative and objective parameters are commonly used to describe image quality: spatial resolution, contrast resolution, temporal resolution, noise, and artifacts. These parameters are affected by CT scanner apparatus and scan variables and are often used to assess the performance of a CT scanner.

Signal An image represents a map of some physical quantity, either directly measured or derived from measurements. The image signal can be continuous, as in a screen-film x-ray or 35-mm photograph, or they can be discrete, such as a medical image on a computer monitor. In the CT acquisition process, the quantity measured is the attenuation of the x-ray beam (just like a projection x-ray), with a continuous physical electrical signal representing x-ray energy flux, converted to a discrete digital value. From a set of these measurements, a digital image is calculated to represent the attenuation coefficient of the material in the object. The map is a collection of pixels (picture elements), typically a square array of 512 pixels on a side. When multiple slices are collected into volume data sets, the 3D map becomes a collection of voxels (volume elements). In computer terms, the original measurements may consist of 16bit data (allowing a range of values spanning a factor of 64,000), whereas the reconstructed images typically are 8or 12-bit data (a range up to 4,095). It is assumed that the signal is linear with the physical properties of the displayed object. For example, if the density of the contrast medium in a voxel doubles, the pixel value will increase by a factor of two. The information in the image signal consists of patterns of change in the image. The magnitude of such change is characterized by contrast, the variation of local values from the surrounding values. In discrete, digital systems, the bit depth of the data determines the smallest change recordable, typically 0.02% step (12 bits) in the digital data or 0.4% step (8 bits) for a displayed image. In the image display process, signal relates to the intensity of light patterns that a human observer views. The dynamic range of light signal may be a factor of 500 to 1,000 from light to dark. Signals can be transformed into different representations, e.g., a CT attenuation image file gets mapped to a light intensity signal for viewing on a monitor, with brightness and contrast adjustments to emphasize different areas of interest.

9

Resolution The term resolution characterizes the ability of an imaging system to detect changes in a signal; the term arises in several different contexts in image operations (e.g., spatial or temporal resolution). The ability of an imaging system to record changes between different points in space depends on two factors: system aperture and (for discrete systems) sampling rate. A system aperture can take different forms: in a display system, it may be the size of the spot of light to form the image; in a CT scanner, it could be the size of the detector cell that measures the x-ray flux. The aperture is considered piece-wise constant within itself, so changes can only be recorded over a size commensurate with its dimensions. Spatial resolution is characterized by a point spread function, which is the signal footprint of an infinitesimal size (point) input, and is expressed as length (such as full-width at half maximum value of the point spread function). Equivalently, the resolution can be reported in the frequency domain by describing a modulation transfer function (MTF), which characterizes how signals of different spatial frequencies (size) are attenuated by the measuring system. In discrete systems, an additional factor affecting resolution is the sampling rate at which signals are transferred. For example, a moving light beam 1 mm in diameter might be modulated every 0.5 mm. Elegant mathematical analyses exist for describing the effect of sampling rate on signal information. One often used result is the Nyquist criterion, which states that at least two samples are required over the distance of the system aperture to prevent distortion of signal information. Such analysis is used extensively in designing medical imaging systems.

Spatial (High-contrast) Resolution Spatial resolution measures the capability of an imaging system to resolve closely placed objects or to display fine details. The spatial resolution of CT is described in two dimensions, xy-image (in-plane) resolution and z-direction (longitudinal) resolution. In-plane and longitudinal resolution depend on different factors. Traditionally the in-plane spatial resolution has been far better than the longitudinal or cross-plane spatial resolution, but the longitudinal resolution has been significantly improved with MDCT and approaches that of the in-plane resolution.

In-Plane Spatial Resolution In-plane spatial resolution is usually expressed in line pairs per millimeter, typically 0.5 to 2 lp/mm for CT. It is often measured directly by imaging and visualizing highcontrast objects of increasingly smaller sizes or increasing spatial frequencies (Fig. 1-8). However, the evaluation process involved in this approach may be subjective. More objective, quantitative methods are based on calculation of MTF, which is defined as the ratio of the output modulation to the input modulation, measuring the response of

5063_Lee_Ch01pp0001-0028 2/17/06 9:38 AM Page 10

10

Chapter 1

Figure 1-8 Bar targets used to determine resolution (from Siemens Medical Systems).

an imaging system to different frequencies. The MTF is most commonly obtained by taking a Fourier transformation of the point spread function that is measured by scanning the cross-section of a thin wire phantom. It is expressed by a plot of the fraction of subject contrast in the image versus spatial frequency. Spatial resolution is then specified at the frequency for a given percent value of the MTF (Fig. 1-9). The spatial resolution of a CT imaging system depends on the quality of raw CT projection data and the reconstruction method. The spatial resolution of the projection data is in turn influenced by system geometric resolution limits such as focal spot size, detector width, and x-ray beam sampling. After a CT scan is acquired, the spatial resolution of the reconstructed CT image can be affected by the selection of a field of view or zoom factor. With the use of a small field of view, the size of the individual pixels decreases and the in-plane spatial resolution of a CT image increases.

Longitudinal Spatial Resolution (Slice Sensitivity Profile) The longitudinal spatial resolution is usually expressed by slice sensitivity profile (SSP),which describes the longitudinal profile of a CT scanner point spread function (Fig. 1-10). The SSP is measured using small thin platelets, just as a thin wire phantom is used to measure the point spread function for the in-plane resolution. The profile is typically a Gaussian curve shape, blurred from the ideal rectangular shape. In spiral CT, the SSP broadens because of the movement of the CT scanner table during the CT gantry rotation. From the SSP, the longitudinal resolution is generally characterized by two numbers: the full width at half maximum or full width at tenth maximum.

The longitudinal resolution has become increasingly important in view of the increasing application of volume examinations and 3D representation. In sequential (SDCT or MDCT) scanning, the SSP is readily defined and mainly determined by x-ray beam collimation. In spiral (SDCT or MDCT) scanning, however, the SSP becomes more complex depending on multiple other factors such as the CT table speed (pitch), spiral interpolation algorithm, detector width (especially in MDCT), and crosstalk between neighboring slices. For spiral SDCT, a higher table speed results in a broader SSP and thus a larger effective slice thickness. The 180LI interpolation algorithm produces thinner SSP than the 360LI algorithm at the cost of higher noise (29,30). For spiral MDCT, the relationship between the table speed (i.e., pitch) and SSP is less straightforward, because multiple sets of spiral data collected from multiple detectors can be interpolated in more complex fashions (25). (See the section titled Computed Tomography Table Travel Speed and Pitch later in this chapter for more detail on the effect of table speed.) With MDCT, variations among CT manufacturers in hardware implementation and algorithmic approaches make it difficult to provide general statements about the effect of scan factors on longitudinal resolution.

Contrast (Low-contrast) Resolution Low-contrast resolution of an image system is the ability to distinguish a low-contrast object from its background. This is a property in which CT is far superior to conventional radiographs. Low-contrast resolution is measured using phantoms that contain objects of varying sizes with small difference in attenuation value from background (Fig. 1-11). The most widely accepted methods of measuring low-contrast resolution of a system are based on the subjective response of an observer to detect objects as distinct. Because the difference between object and background signal is small, noise plays an important role in determining low-contrast resolution. Many factors that influence the noise level such as tube current, tube voltage, slice thickness, and reconstruction algorithm also affect low-contrast resolution. In addition to these factors, the size of objects and viewing window setting also affect lowcontrast detectability. In CT, the contrast difference between objects is typically characterized by the percentage linear attenuation coefficient: 1% contrast difference corresponds to a difference of 10 HU.

Temporal Resolution Temporal resolution determines how rapidly changing signals can be recorded. As CT technology advances, temporal resolution increases continuously with an increase in volume coverage and scan speed. High temporal resolution is particularly desirable when imaging moving structures (e.g., heart, lungs) and for dynamic contrast medium

5063_Lee_Ch01pp0001-0028 2/17/06 9:38 AM Page 11

Basic Principles of Computed Tomography Physics and Technical Considerations

11

1

0.5

0

50

100

150

200

250

300

350

400

450

500

50

100

150

200

250

300

350

400

450

500

50

100

150

200

250

300

350

400

450

500

0.04

0.02

0 0.03 0.02 0.01

A

0

Figure 1-9 The point spread function describes how an impulse signal becomes broadened by the system aperture. A: Three different apertures sample an impulse train, spreading the signal over a wider area. Impulses that are spaced close together cannot be resolved by wider apertures. An equivalent description can be given in the frequency domain by the modulation transfer function. Rectangular apertures acting on a sine wave diminish the peaks of the measured sine wave—a plot of the modulation versus frequency of the sine waves gives the modulation transfer function. (continued)

enhancement applications (e.g., CT angiography, perfusion studies).

Distortion A signal can contain errors or distortions that are repeatable (deterministic). For instance, if a patient moves during the acquisition step, parts of the anatomy may be blurred or in different positions from their true location. If the image reconstruction process requires linear and consistent data, but the measurements for some reason are not consistent, artifacts can arise that may result in lost information or spurious image features. Because distortion is deterministic, in some cases it is possible to remove or minimize the error by additional image processing. For example, if a particular set of pixels is mapped to brightness larger than a cathode-ray tube can produce, the window/level can be adjusted to allow the pixels to display at a lower brightness. Because such artifacts manifest themselves in ways unique to specific imaging tasks, there is no general methodology for describing artifacts universally.

Noise Another class of errors in images is due to random, stochastic variations in the signal, which are not repeatable. Because noise cannot be predicted, its effects cannot be corrected, and it becomes the ultimate limitation in the amount of information that can be extracted from a signal. Noise can be characterized by its statistical properties, such as the variance or standard deviation of repeated measurements. The measure of most interest is not the absolute magnitude of these properties (which can change arbitrarily as the signal is transformed by different operations), but rather the ratio of the signal mean to its standard deviation, or signal-to-noise ratio (SNR). It is the SNR that determines the amount of information that can be derived from a measurement. Many image types (such as x-rays or photography) are formed by photons, which have noise properties described by Poisson statistics. The mean and noise variance in this case have a very simple relationship (the variance equals the mean), and hence the square of the SNR is simply equal to the mean number of quanta. This quantity is called noise equivalent quanta (NEQ) and

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 12

12

Chapter 1 1 0.9 0.8 0.7 0.6 0.5 0.4 0 1 0.8 0.6 0.4 0.2 0 −0.2 −0.4 −0.6 −0.8 −1 0 1 0.8 0.6 0.4 0.2 0 −0.2 −0.4 −0.6 −0.8 −1 0 1 0.8 0.6 0.4 0.2 0 −0.2 −0.4 −0.6 −0.8 −1 0

0.1

0.2

100

100

100

0.3

200

200

200

0.4

300

300

300

0.5

0.6

400

400

400

0.7

500

500

500

1 0.8 0.6 0.4 0.2 0 −0.2 −0.4 −0.6 −0.8 −1 0 1 0.8 0.6 0.4 0.2 0 −0.2 −0.4 −0.6 −0.8 −1 1 0.8 0.6 0.4 0.2 0 −0.2 −0.4 −0.6 −0.8 −1

100

200

300

400

500

0

50 100 150 200 250 300 350 400 450 500

0

50 100 150 200 250 300 350 400 450 500 B

Figure 1-9 (continued) B: The modulation transfer function (top) is plotted, showing a decrease in amplitude with increasing frequency. Below it are plotted on the left input sine waves (increasing with frequency from top to bottom) and on the right are output sine waves, demonstrating amplitude reduction at higher frequencies.

is a widely accepted metric for characterizing noise properties in an image (5). In clinical practice, CT values are often measured to characterize the composition of tissues such as air, fat, water, and calcification. Measured CT values are subject to inherent system physical limitation and statistical fluctuation (or “noise”). Image noise also reduces low-contrast detectability. Image noise is often simply measured as the standard deviation of voxel values in a homogeneous water phantom (Fig. 1-12). Image noise is affected by a large number of parameters such as x-tube voltage, tube current, exposure time, detector

efficiency, slice thickness, table speed, interpolation, and reconstruction algorithm. Image noise can be reduced with an increase in x-tube voltage and tube current, but at the expense of a higher radiation dose. Conversely, reduction in radiation dose may increase noise and impair lowcontrast resolution. For example, reduction of the milliamperes by one half increases noise by 41%. An increase in slice thickness reduces noise but at the expense of a decrease in the longitudinal resolution. A smooth reconstruction algorithm reduces noise but decreases in-plane spatial resolution. These tradeoffs among imaging parameters should be understood in setting up scanning and image

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 13

Basic Principles of Computed Tomography Physics and Technical Considerations

13

1.5 1 0.5 0 −50

0

50

100

150

50

100

150

200

1.5 1 0.5 0 0 1

0.5

Figure 1-10 Slice sensitivity pro0 100

150

200

reconstruction protocols. In single-detector row spiral CT the 360LI interpolation algorithm may be used instead of the 180LI when noise reduction is more desirable than longitudinal spatial resolution. This simple relationship, however, may not be applicable in MDCT.

IMAGE VISUALIZATION AND ANALYSIS Traditionally, body CT image review has been limited primarily to the transverse plane. MDCT, however, facilitates

250

300

files: (top) ideal square response; (middle) penumbra broadens edges; (bottom) travel blurs response.

rapid acquisition of volumes of image data with nearly isotropic resolution. With the increasing capability of computers to process large data sets rapidly, clinicians can readily reformat the transaxial images into arbitrary orientations, using either 2D or 3D techniques, to achieve appropriate views. Five common rendering techniques are currently in use: 2D multiplanar reformations (MPR), 3D maximum-intensity projections (MIP) or minimum-intensity projections (MINIP), 3D shaded-surface display (SSD), 3D volume-rendering display (VRD) and perspective rendering. All 3D displays are based on rendering volumetric CT dataset to generate 2D images that are oriented normal to a specific viewing point.

Multiplanar Reformation (Two-Dimensional)

Figure 1-11 Contrast detail objects test low contrast resolution. Noise obscures small differences.

MPR is the simplest and most commonly performed data representation. The volumetric CT dataset that is acquired from a series of contiguous transverse images can be thought of as a 3D stack of individual voxels. This dataset can be reconstructed as sections in any desired plane, generating 2D MPR images. Three orthogonal planes are commonly selected, resulting in transaxial, coronal, and sagittal MPRs (Fig. 1-13). The planes can be rotated at any arbitrary angle to generate oblique MPRs. With a curved line drawn manually or automatically, the data also can be visualized along a curved plane (curved MPR) (Fig. 1-14). Although typically a 2D MPR generated in the selected plane is a single voxel deep, arbitrarily thick sections (MPR

5063_Lee_Ch01pp0001-0028 2/17/06 9:38 AM Page 14

14

Chapter 1

A,B

C

D,E

F Figure 1-12 Computed tomography simulations of the abdomen at the midportion of the kidneys in a 2-year-old child. A: 0.5% of the radiation dose; B: 5% of the radiation dose; C: 15% of the radiation dose; D: 30% of the radiation dose; E: 60% of the radiation dose; F: original, full-dose image. There are two subtle lesions in the right kidney, the larger lesion (arrow in F) can be detected to 15% of the radiation dose. The smaller lesion (arrowhead in F ) can be detected to 30% of the radiation dose. (Courtesy of Dr. Steven Don.)

slabs) can be readily produced by averaging the voxels within the slabs. MPR is easy to generate and very effective for displaying anatomic features that are oriented perpendicular to the x, y plane. Curved MPR is useful to delineate curved or mildly tortuous structures that are not contained within a single plane. Limitations of the curved MPR technique are that represented anatomy may appear distorted and difficult to evaluate, and that an inaccurate selection of the curve on the reference image can lead to misrepresentation of the structures being evaluated. The interpretation of such images should be carefully exercised in conjunction with the source transaxial or standard MPR images.

Maximum-Intensity or Minimum-Intensity Projections (Three-Dimensional) MIP is generated by creating an image plane consisting of the highest voxel values along each of a series of parallel rays projected normal to the viewing point. This technique is widely used when structures of interest have higher attenuation than adjacent structures such as skeletal structures

and contrast medium–enhanced vessels (Fig. 1-15). Unlike other volume rendering techniques, the relative density information (i.e., voxels) with the highest attenuation is preserved. However, MIP does not provide depth cues. To reduce this lack of spatial relationship, a series of MIPs is usually generated and displayed by rotating the viewing points at small increments. Instead of processing the entire volume, the volume of interest can be confined to generate a series of MIP images in thin slabs without including extraneous bright objects (e.g., sliding thin-slab MIP). The thickness of the slab can be adjusted by the user. Similar to MIP, MINIP can be generated to display the structures with lowest CT values or air attenuation, such as bronchi, pulmonary emphysema, and colon (Fig. 1-16).

Shaded-Surface Display and Volume-Rendering Display (Three-Dimensional) SSD is one of the early 3D display techniques and was popular when computers had a limited capability to process a large amount of data. In this technique, a subset

5063_Lee_Ch01pp0001-0028 2/17/06 9:38 AM Page 15

Basic Principles of Computed Tomography Physics and Technical Considerations

15

A

B

Figure 1-13 Multiplanar representation images displayed in the (A) transaxial, (B) oblique sagittal, and (C) coronal planes. Figures 1-13A and 1-13B demonstrate multifocal transitional cell cancer involving the bladder wall (black arrows) and distal left ureter (white arrows) in the same patient. Duplicated right ureters are clearly depicted in C, which is a thick-slab multiplanar representation image from a different patient.

of volumetric data is retained for surface rendering and the remainder is removed. The subset can be determined by including voxels in a range of attenuation values or by allowing the computer to connect surface voxels automatically from a user defined region or seed point. The segmented surface voxels are modeled as a collection of polygons and displayed with shading to provide depth cues. Although the SSD may be computationally advantageous, it is not commonly used in clinical practice because of two fundamental limitations: the dependence of accurate display on the predefined threshold value and the lack of visualization of internal structure. With the availability of increasingly powerful computers, SSD has been largely replace with VRD, which has the advantage of using and displaying the entire volume dataset. The VRD is well suited for a wide range of visualization tasks because of its flexibility (Fig. 1-17). The contributions

C

from each voxel are integrated and can be readily displayed in different perspectives. By changing parameters, the data can be displayed with varying degrees of opacity, shading, brightness, and color to meet the demands of a specific clinical application. In commercial display workstations, these parameters are often optimized and implemented as preset buttons that allow the user to readily choose a desired display task or an organ of interest.

Perspective-Rendering Display (Three Dimensional) The images in VRD are commonly rendered with the selection of an observer position external to the data set. Perspective rendering represents a special form of VRD or SSD in which the images are generated with divergent rays to simulate the perspective of endoscopy. In this display, structures closer to the observer position appear larger than those farther away, resulting in virtual endoscopy. The

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 16

16

Chapter 1

Figure 1-14 Curved multiplanar representation images of the left anterior descending artery (LAD) are shown on the two right columns. On the left image, the curve trajectory is displayed superimposed over the LAD of the reference volume rendered heart. The five individual square images in the left column represent cross-sectional images of the LAD perpendicular to the curve.

perspective rendering display has been applied to anatomic structures that are amenable to endoscopy, such as gastrointestinal tract, airway, and blood vessels (Fig. 1-18). The color and opacity functions are selected to delineate the lumen wall while keeping the internal structure transparent. Navigation along the course of the selected anatomy of interest can be presented as “fly-through,” which mimics a virtual flight. Although virtual endoscopy display requires more user interaction and image processing than other display techniques, the development of automated or advanced tools such as automated center path tracking, unraveling, and multiperspective visualization may make this technique more practical in clinical practice.

Image Analysis and Computer-Aided Diagnosis Computer analysis of radiographic images has been practiced for many years. Its application, however, has been mainly limited to research, with infrequent usage for routine interpretation of clinical images. With recent advances in computer and image processing techniques, image analysis tools now are readily available and have become a standard feature available on many image reading workstations. Image measurement tools vary a great deal in their image analysis capability, ranging from a simple attenuation measurement of a region of interest to fully automated computer-aided diagnosis (CAD) techniques. The detailed description of various image processing and analysis methods and associated clinical applications is beyond of the scope of this chapter. Some common applications include attenuation measurement of a lesion, histogram analysis of a region, and length and size

measurements of anatomic structures. For example, a region of interest is placed over an adrenal mass to measure its mean attenuation value. A mean threshold of 10 HU or less is commonly used to assess the presence of lipid pixels that suggest the diagnosis of adrenal adenoma. Similarly, the attenuation difference before and after administration of contrast medium is measured to evaluate contrast enhancement in a mass. An increase in attenuation value of more than 15 to 20 HU is considered to be indicative of contrast enhancement within the mass. Other quantitative applications include coronary artery calcification measurement, attenuation histogram-analysis based pulmonary emphysema measurement, and bone mineral density measurement. The clinician must be aware, however that CT attenuation measurements are subject to variation among scanners, particularly scanners of different manufacturers. Therefore, CT attenuation measurements may not be reliable without a rigorous quality control and standardized measurement procedure. With the 3D volumetric imaging capability of MDCT, CT images are routinely used for morphometric analysis of anatomic structures. For example, in the application of endoluminal graft intervention for an aortic aneurysm, multiple measurements including the luminal and aneurysm diameters and lengths of the aorta and other large vessels are often obtained to assess the technical feasibility of the intervention and to select appropriate graft designs. In the 1970s, about the time when CT was introduced, artificial intelligence research was launched to exploit the computational power of a machine to solve various problems. Expert system software was developed to simulate the performance of experts in many fields, including that of medical diagnosis. As an expert system, CAD technology uses powerful image-processing algorithms to segment and extract geometric and shape features from diagnostic images. The features are further processed and synthesized for the detection and classification of abnormalities. CAD has been steadily expanded and improved with the advance of computer and imaging technology. In particular, CAD has been widely accepted as an aid in detecting breast masses and microcalcifications in mammography. Although it has yet to be implemented on a routine basis for clinical CT image evaluation, CAD has shown promising results for lung nodule volume measurement, automated lung nodule detection, and colonic polyp detection.

COMPUTED TOMOGRAPHY SCAN PARAMETERS Collimation The x-ray beam emitted from the focal point on the anode plate in the x-ray tube is collimated to reduce unwanted radiation exposure, generate desirable image slice profiles, and guard the detector against scattered radiation. Collimators may have variable configurations depending on

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 17

Basic Principles of Computed Tomography Physics and Technical Considerations

17

Figure 1-15 Maximum-intensity projection im-

A

B

the type of scanner, but their basic functions are the same. Collimators are located at two points within the CT gantry: near the x-ray tube (prepatient collimator) and near the detector (postpatient collimator). The prepatient collimator is used to shape the x-ray beam. It has two components. The first component is a fixed collimator that defines the fan shape of the x-ray beam in the transverse plane. The second component is an adjustable collimator that allows variable collimation to the desired beam thickness in the longitudinal plane. This beam thickness is the collimation size commonly referred to in clinical protocols. For example, for a 64  0.5-mm detector configuration in a 64-MDCT, the beam collimation is 32 mm, whereas for a 16  1-mm detector configuration

ages of (A) peripheral run-off computed tomography angiography and (B) computed tomography urography in the anteroposterior plane. Figure 115A demonstrates an acute segmental obstruction in the right distal superior femoral and right proximal popliteal arteries (arrows) and some vascular calcifications (arrowhead ). The bone was segmented out prior to the generation of the maximum-intensity projection image in this figure. B delineates normal anatomy of contrast-enhanced kidneys, collection systems, ureters, and bladder in computed tomography urography along with normal bony structure.

in a 16-MDCT, the beam collimation is 16 mm. (These dimensions are defined at the point where the fan beam crosses the isocenter.) Postpatient collimators are located in front of the detector and serve to reduce scattered radiation and to improve the SSP. An ideal point focus would result in a perfect rectangular slice profile (Fig. 1-19). Any real finite focus, however, yields a collimated beam profile shaped like a trapezoid because of the penumbra effect. The umbra is the region in the detector illuminated by the x-ray beam emitted from the entire area of the focal spot and corresponds to the plateau region of the trapezoid. The penumbra is the region illuminated by the x-ray beam emitted from only a part of the focal spot and corresponds to the

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 18

18

Chapter 1

A

B

edge of the trapezoid peripheral to the plateau. In SDCT, the entire trapezoid collimated x-ray beam can contribute to the detector signal, and the slice width is estimated as the full-width half maximum of the trapezoid. In MDCT, however, only the umbral region that provides an equal signal level for all detector elements is used, because the penumbral region has spatially varying x-ray intensity. The penumbral region is discarded from the data acquisition and thus contributes to the radiation dose in MDCT being higher than that in SDCT. The relative contribution of the penumbral region decreases with increased beam thickness, thereby making 16-MDCT superior to 4-MDCT in utilization of the collimated beam.

Computed Tomography Table Travel Speed and Pitch In spiral scan mode, CT table travel speed is an important parameter that is closely associated with image quality, radiation dose, scan time and volume of coverage. MDCT and wide x-ray beam coverage facilitate faster CT table travel per gantry rotation. Pitch is a parameter that is commonly used to characterize the CT table movement. It is defined as the travel distance of the CT scan table per 360-degree rotation of the x-tube divided by the x-ray beam collimation width (30). When the table feed and the beam collimation are identical, pitch is 1. When the table feed is less than the beam collimation, pitch is less

Figure 1-16 Minimum-intensity projection images of (A) normal airway and (B) stenotic trachea and occluded right main stem bronchus (arrows).

than 1 and scan overlap occurs. Scan overlap increases as pitch decreases. With 4-channel MDCT at 4  1.0-mm collimation and a table-feed of 6 mm per rotation, the pitch is 6/(4  1)  6/4  1.5. With 64-channel MDCT at 64  0.5-mm beam collimation and a table-feed of 48 mm per rotation, the pitch again is 48/(64  0.5)  48/32  1.5. It should be noted that in MDCT scanners pitch is sometimes defined by using the detector aperture as the denominator instead of x-ray beam width. A pitch value in this pitch definition (termed detector-pitch) corresponds to an original-definition of pitch value (termed beam-pitch) multiplied by the number of detector rows: for example, detector-pitch 16 corresponds to beam-pitch 1 in 16-channel MDCT. The pitch represented in the detector-pitch definition does not demonstrate clearly the degree of overlap between adjacent scans and is less useful for radiation dose and image quality consideration. Thus, the original beampitch definition should be used. The impact of pitch is more pronounced in SDCT than MDCT, in part because use of a high pitch is crucial to compensate for the limited scan coverage of SDCT. Highpitch values in SDCT allow faster scans and reduce patient radiation dose but broaden the SSP. Notably, the image noise in SDCT is independent of the pitch if the tube current remains constant throughout the scan; at high pitch values, the section is broadened but the number of photons per section is unchanged.

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 19

Basic Principles of Computed Tomography Physics and Technical Considerations

A

19

B

C

D Figure 1-17 Volume rendered images of (A) computed tomography urography, (B) peripheral computed tomography angiography, (C) cardiac computed tomography angiography, and (D) computed tomography colonography. A: demonstrates duplicated right ureters. B: delineates extensive vascular calcifications and multiple foci of stenoses in bilateral lower extremity arteries. C: shows three coronary arteries that were segmented and colored distinct from the cardiac chambers. D: shows a transparent or hollow representation of air-filled colon and small bowel loops delineated in a computed tomography colonography.

5063_Lee_Ch01pp0001-0028 11/01/05 1:47 PM Page 20

20

Chapter 1

pitch values from 1 to 2 are commonly used in both SDCT and MDCT, except in cardiac applications in which lowpitch overlapping scanning is acquired to ensure sufficient scan data sampling contiguously over the volume without a gap in the coverage. Furthermore, low pitch value scanning is more effective in reducing “windmill” artifact, which is associated with MDCT and is more apparent in MPR or 3D image display.

Tube Voltage and Current

Figure 1-18 Endoscopic or perspective rendering representation of a small colonic polyp and colonic wall from a computed tomography colonography.

Pitch has a smaller effect on image quality in MDCT than in SDCT, but its association with image quality, artifact, and radiation dose is complex and controversial. Optimal choice of pitch may depend on the detector configuration and CT projection data interpolation schemes. Some manufacturers recommend the use of a set of fixed pitch values in MDCT, whereas others advocate flexible selection of pitch values. In general, with a high pitch, the longitudinal resolution is degraded because of the broadening of the section profile. With a low pitch, the longitudinal resolution improves, but a higher radiation dose is required to maintain the signal-to-noise ratio. This tradeoff between the image quality and radiation dose should be considered when a pitch value is selected for the scan protocol in a particular clinical application. In practice,

Figure 1-19 Perfect rectangular slice profile. Modified from Kalender W. Computed tomography. Munich: Publicis MCD Verlag, 2000.

Appropriate selection of CT scan parameters is critical to optimize radiation exposure and image quality. Reduction of the tube voltage at a constant tube current or reduction of the tube current at a constant tube voltage decreases the output of the x-ray tube and radiation dose to the patient. However, an inappropriate reduction of the voltage may result in a marked increase in CT tissue attenuation and noise, particularly in large patients. In most CT scanners, there are only a few choices of kVp selection for CT scans. Routine body CT for adult patients is performed with 120 to 140 kVp. Scanning pediatric patients with 80 kVp has been well accepted for reducing radiation. An additional consideration in selecting kVp is that the CT attenuation by iodine atoms, used in contrast mediumenhanced studies such as CT angiography, increases as kVp approaches 80, when the effective photon energy (approximately one half of the kVp) is closer to the K-edge of iodine (i.e., 33.2 keV). For example, the contrast medium concentration that results in contrast enhancement of 250 HU at 120 kVp will yield contrast enhancement of 400 HU at 80 kVp. In practice, however, the maximum x-ray tube current available at 80 kVp is in general insufficient to scan large patients or large body parts such as the abdomen and pelvis in adults. Furthermore, the x-ray absorption is higher at lower photon energies and may contribute to higher effective radiation doses. Compared with kVp, the choice of the x-ray tube current is more flexible, typically ranging from 20 mA to 800 mA. Another practical advantage of adjusting the tube current instead of tube voltage is that its effect on the image quality is more straightforward. Thus, manipulating the tube current or rotation time instead of the tube voltage is a more common, practical approach to modifying radiation dose. For example, in chest CT, a protocol for pulmonary nodule screening may use 20 mAs at 120 kVp, whereas a routine clinical chest protocol uses 120 mAs at 120 kVp. In SDCT, higher pitch values result in thicker slices, and thus, at a constant tube current, noise per slice remains constant. With MDCT, however, increasing the pitch does not necessarily generate thicker slices. When the slice thickness is unchanged, increasing pitch at a constant tube current lowers the radiation dose and increases noise in an image. To keep the noise level constant, the

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 21

Basic Principles of Computed Tomography Physics and Technical Considerations

tube current must be increased at the higher pitch. Thus, a new term effective mAs was introduced, which is defined as the ratio of the mAs divided by the pitch. A scan with 200 mAs at pitch 2 has the same effective mAs as a scan with 100 mAs at pitch 1, which make the two scans dose and noise equivalent.

Reconstruction Kernel Various filters may be used in filtered back projection reconstruction. The filtering is done by convolution kernels (or reconstruction algorithms), which trade off sharpness of image features versus the introduction of background noise. This tradeoff stems from the fact that the random fluctuation of x-ray flux occurs at a very small scale and hence has high frequency content, whereas the signal response is limited by the frequency of the detector aperture. The enhancement of signal is typically less than the noise enhancement, and hence the SNR falls as more detail is presented. High-resolution reconstruction kernels or algorithms, such as bone or lung kernels, produce higher spatial resolution but with increased noise in the image. Lowresolution kernels, such as soft tissue or smooth kernels, decrease image noise but also decrease spatial resolution. In image reconstruction, kernels should be selected to give an appearance appropriate for the clinical task (Fig. 1-20). Some of newer scanners routinely generate images with different kernels, such as soft tissue and lung kernels in chest CT scan, as a standard protocol on completion of a single acquisition of projection data.

21

Sequential and Spiral Scan Modes With the advances of spiral and MDCT, spiral scanning has become the standard CT scan mode. Sequential (step-andshoot) scan mode continues to have a few clinical applications such as contrast bolus monitoring CT, perfusion CT, interventional applications, and electrocardiography (ECG)-triggered coronary artery calcium CT, in which scans are either repeated at the same location or acquired at a delay interval between CT table positions. The number of images acquired in sequential scanning depends on the number of active detector sections (or channels). By combining the detector signals of the individual sections during image reconstruction, the number of images per scan can be reduced, and the image section width can be increased. As an example, a scan with 16 sections at 0.5-mm collimation provides either 16 images with 0.5-mm section thickness, eight images with 1.0-mm section thickness, or two images with 4.0-mm section thickness. This trade-off between the number and thickness of images is equally applicable to spiral scanning.

Section Thickness The choice of section thickness depends on both the clinical application and the quantitative or display requirements. Images with thin sections provide anatomic details but with increased amounts of data and time required for image interpretation. Furthermore, images with thinner sections typically take longer to acquire than the images with thicker sections and are noisier. A commonly used

A

B Figure 1-20 Chest image with soft tissue (A) and bone kernels (B).

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 22

22

Chapter 1

slice thickness for routine diagnostic applications is 5 mm. Images for 3D display, CT angiography, or detection of small pulmonary nodules are usually acquired with a 1- to 2-mm section thickness. Some clinical applications associated with quantification of fine structures (e.g., small plaques in the coronary artery or temporal bone anatomy) may require 0.4- to 0.6-mm section thickness (Fig. 1-21). In sequential scan model, the section thickness or longitudinal resolution of images is determined by x-ray beam collimation. For spiral SDCT, the section thickness is broadened with higher pitch values and with the 360LI reconstruction algorithm compared with the 180LI algorithm. Because of this broadening, pitch is usually limited to less than 2. Although narrow beam collimation with high pitch may provide the same scan coverage as broad

beam collimation with low pitch, the former is preferred to the latter for the reduction of section thickness. With spiral SDCT, scan speed is limited. To improve scan coverage with SDCT, section thickness can be increased by using thick collimation and high pitch. To compensate for the increased section thickness, a small reconstruction interval can be used to produce overlapped image sections. CT imaging with overlapping reconstruction enhances the 3D representation of volumetric data and provides images with better image quality than images that are interpolated by a computer image-processing algorithm from CT images after reconstruction. Overlapping reconstruction also improves the detection of small lesions by increasing the number of images to review and the likelihood that the center of the lesions, with maximal

A

B

Figure 1-21 Image with differing section thicknesses: 1 mm (A), C

3 mm (B), and 5 mm (C).

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 23

Basic Principles of Computed Tomography Physics and Technical Considerations

contrast, is visualized in the images. Stair-step artifact is also reduced with smaller section size. In SDCT, section thickness is determined and fixed with the acquisition of scan projection data. In contrast, with MDCT, spiral data acquired during a single CT gantry rotation can generate CT images with different section thicknesses. The section thickness, however, cannot be decreased below the collimated detector width used during acquisition. For example, from a scan acquired with 16  0.5-mm collimation in 16-channel MDCT, images can be produced with varying section thicknesses of 0.5, 1, 1.5, 2, 3, 4, 5 mm, and so forth. With the use of larger section thickness, the number of images that can be reconstructed is reduced and the noise per image is less. The increased flexibility of image reconstruction with MDCT has improved the efficiency of clinical applications. For example, when a chest CT scan is performed with narrow collimation, thick-section images are generated initially and reviewed for diagnostic purposes. If thinner section images are desired to better characterize a nodule, they can be reconstructed readily from the projection data. Thinsection images can be generated from the same CT scan projection data for 3D visualization and CT angiography. This option of generating a wider section by summing several thin sections is beneficial for examinations that require narrow collimation to prevent partial volume artifacts. For example, in the head, partial volume artifacts of dark streaks or areas of hypoattenuation can be markedly reduced with the use of thinner collimation. One disadvantage of using thin collimation is decreased scan coverage, although this becomes less critical the higher the number of detector rows.

RADIATION DOSE CT scanning constitutes the largest single medical-related radiation exposure for the general population (26). With the increasing availability of modern CT scanners and their expanding clinical applications, radiation exposure to the patient at CT is a potential serious clinical concern. Although insufficient radiation dose results in increased noise and degradation of image quality, an increase in radiation dose above a certain level may not contribute to improving diagnostic image quality and merely deposits more radiation into the patient’s body. Each CT scan protocol should be carefully assessed and designed to limit radiation exposure. The development of practical guidelines and the application of appropriate strategies based on clinical indications and technical innovations are crucial to minimize radiation dose.

Fundamental Radiation Dose Measurement Radiation is quantified in several different units, making it difficult to compare doses for specific CT applications re-

23

ported in the published literature. Radiation exposure is the most basic quantity in radiation dosimetry and is related to the amount of ionization produced by an x-ray beam per mass of air. It is measured in coulombs per kilogram or unit of roentgens (1 R  2.58  104 C/kg). As a result of exposure, the amount of radiation dose absorbed in the patient’s body is expressed in rad or gray (1 rad  10 mGy). It should be noted that the term radiation exposure is a radiation source-related term and is a measured quantity, whereas radiation dose is a body-related term and must be calculated from the exposure by applying a conversion factor. Unfortunately, the terms radiation exposure and radiation dose are often used interchangeably. The conversion factor for calculating radiation dose from the radiation exposure depends on both the absorbing material (e.g., air, soft tissue, bone) and its position in the object exposed to radiation. Absorbed radiation dose does not account for differing sensitivities of the organs to radiation damage. Therefore, equivalent or effective dose in a tissue is a product of the tissue type and the radiation weighting factor. The weighting factor is approximately unity for x-rays, so the equivalent dose has the same numerical value as absorbed dose and is measured in millisievert or rem (10 mSv  1 rem). Effective dose is computed by summing the absorbed doses in individual organs weighted by their radiation sensitivity. Effective dose estimates the whole-body dose that would be required to produce the same risk as a partial-body dose delivered by a localized radiological procedure. Effective dose is useful in evaluating and comparing the potential biologic risk of a specific radiologic examination. It allows comparison of the measurement with other types of whole-body radiation exposure such as that due to natural background radiation.

Radiation Dosimetry Parameters Specific to Computed Tomography Various CT radiation dosimetry parameters and the approach to their measurement are described graphically in Figure 1-22. The basic radiation dose parameter in CT is the CT dose index (CTDI), which represents absorbed radiation dose in a CT-dose phantom and is measured in gray (Gy) or rad. There are three derivatives of the CTDI: CTDI100, CTDIw, and CTDIvol. The CTDI100 represents radiation exposure measured by means of an ionization chamber with a length of 100 mm. The ionization chamber is placed in a cylindrical polymethylmethacrylate (Plexiglas) head (16-cm diameter) or body (32-cm diameter) phantom to measure radiation exposure of a single axial scan over a length of 100 mm (Fig. 1-23). Because radiation exposure measurements at the center and the periphery of the phantom are different, the weighted average of CTDI100 is calculated by adding one third of the center values and two thirds of the periphery values of CTDI100. This weighted exposure measurement is converted to weighted

5063_Lee_Ch01pp0001-0028 2/17/06 9:39 AM Page 24

24

Chapter 1

(1) Radiation exposure (C/kg, R): measured with phantom, CTDI100

(2) Absorbed radiation dose (Gy, rad): CTDIw = (2/3 central + 1/3 peripheral CTDI100) x 33.7 CTDIvol = CTDIw /pitch (i.e., normalized by pitch)

(3) Accumulative radiation dose (mGy x cm, mrad x cm): DLP = CTDIvol x scan length

(4) Effective or equivalent dose (mSv, rem): E (mSv) = DLP x 0.017

radiation dose (CTDIw) by multiplying the adsorption factor (33.7 Gy/C/kg or 0.87 rad /R). In clinical practice, a volume of interest (equivalent to multiple adjacent sections) is scanned, rather than just a single section. The dose for a particular section is increased due to the contributions from adjacent sections. The cumulative or volumetric radiation dose is directly related to the spatial separation of successive scans. To account for this overlapping section effect, the volume CTDI (CTDIvol) was introduced and derived from the CTDIw. The overlap between successive scans depends on the table motion

Figure 1-23 Instruments for computed tomography radiation exposure. Plexiglas computed tomography body (black arrow) and head phantoms (white arrow) are placed over a computed tomography table. An ionization chamber probe (arrowhead) is inserted in the center of the body phantom to measure the central computed tomography dose index of 100.

Figure 1-22 Various computed tomography radiation dosimetry parameters and the approach to their measurement. Bae KT, Whiting B. Radiation dose in multidetector row computed tomography cardiac imaging. J Magn Reson Imaging 2004;19(6):859–863.

during the scan and is described by pitch in spiral CT. When the table feed is less than the beam collimation, pitch is less than 1 and scan overlap occurs. Scan overlap increases as pitch decreases. Overlapping sections or spiral scans with a pitch of less than 1 result in a higher volume CTDI than non-overlapping sections. Therefore, the volume CTDI (CTDIvol) corresponds to CTDIw/pitch. The measurement unit of CTDIvol is Gy. The CTDIvol is now the preferred measure of radiation dose in CT. It is readily displayed in most current CT scanners and allows comparison of the radiation doses among different imaging protocols. The CTDIvol, however, does not take into account the length of the scan or the total number of successive scans. To overcome this limitation, the dose-length product (DLP) was introduced (DLP
CTDIvol  scan length). The DLP represents the integrated radiation dose for a specific CT examination and is expressed in mGy-cm. From DLP, effective dose for typical regions of the body can be calculated by applying a conversion factor (mSv/mGy/cm) that depends on organ risk-weighting factors. Effective dose for a CT scan can be estimated by using dose distributions for specific CT scanner geometry and beam quality, which in turn is related to tube current, voltage, scanned volume, and pitch (1,24,33). Typical effective doses for common radiologic examinations and for background radiation are listed in Table 1-1 (24).

Computed Tomography Scan Parameters Affecting Radiation Dose Radiation dose can be modified by adjusting the tube voltage, tube current, pitch, scan time, and scan range. As discussed earlier in this chapter, reduction of the tube voltage

5063_Lee_Ch01pp0001-0028 2/17/06 9:39 AM Page 25

Basic Principles of Computed Tomography Physics and Technical Considerations Single-slice CT

TABLE 1-1

25

Multi-slice CT

focus

COMPARISON OF EFFECTIVE RADIATION DOSES∗ Procedure

Effective Dose (mSv)

Posteroanterior chest radiograph Head computed tomography Chest computed tomography Abdomen and pelvis computed tomography Diagnostic coronary angiogram Annual natural background radiation ∗Summarized

collimator

0.05 2–4 5–7 8 – 11 3–6 2.5 – 3.6

axis of rotation

8 mm slice incl. penumbra

4x2 mm slices excl. penumbra

from references (3,15,22,27,28).

at a constant tube current or reduction of the tube current at a constant tube voltage decreases the output of the x-ray tube and thus the radiation dose to the patient. Instead of manipulating the tube voltage, a more common, practical approach of modifying radiation dose is to adjust the tube current or rotation time. Another practical advantage of adjusting the tube current instead of tube voltage is that its effect on the image quality is more straightforward. Radiation dose and image noise are affected by the product of tube current and gantry rotation time. At constant image noise levels, the radiation dose delivered at 120 kVp with a relatively higher mAs value is similar to that delivered at 140 kVp with a relatively lower mAs value. Thus, in practice, the choice of kVp-mAs combination can be flexible and up to the individual radiologist. Radiation dose in spiral CT is affected by pitch, as discussed earlier in this chapter. For SDCT scanners, radiation dose and scan duration decrease proportionally with increasing pitch if the tube voltage and current are kept constant. The drawback of high-pitch SDCT is that the SSP broadens and partial volume averaging increases with increasing pitch. For MDCT scanners, the relationship between pitch and radiation dose is not always linear. The tube current is frequently increased to compensate for increased noise at higher pitch values, and thus increasing pitch is not directly translated into reduction in radiation dose (23). At constant image noise levels, the use of effective tube current results in an effective dose independent of pitch. Multiple physical aspects of CT scanners contribute to wasted radiation dose, and the efficiency of a scanner in minimizing wasted radiation is collectively described as the geometric efficiency of the CT scanner (10,23,36). MDT scanners in general have lower geometric efficiency than SDCT scanners because of the gaps between detector elements in the detector array and the use of wider beam width. In SDCT scanners, radiation in the x-ray beam penumbra is used in the formation of the image. In MDCT scanners, however, use of this portion of the beam would result in inconsistent beam intensity measurements com-

detector 1 “24 mm” element fully and partly exposed

8 “1 mm” elements fully exposed

Figure 1-24 In single-detector row computed tomography scanners, radiation in the penumbra that exposes the patient is used in the formation of the scan data. In multidetector computed tomography scanners, however, this portion of the beam would result in beam intensity measure that is not as consistent as the central or umbra portion of the beam discussed in the text. Modified from Kalender W. Computed tomography. Munich: Publicis MCD Verlag, 2000.

pared with the central or umbra portion of the beam (as discussed earlier in this chapter) (Fig. 1-24). Therefore the “overbeaming” or penumbra radiation of MDCT scanners does not contribute to image generation but does contribute to patient radiation dose. The wider the collimated beam is, the smaller the percentage of radiation that is wasted due to the penumbra. This effect is greatest on four-channel MDCT scanners that operate with narrow collimation modes and lessens progressively as the number of detector rows increases because the fractional effect of the penumbra per detector row becomes proportionally less.

Approaches to Reduce Radiation Dose Careful selection of CT scan parameters is imperative to optimize imaging protocols such that diagnostic-quality CT images are generated with the least radiation exposure to the patient, satisfying the ALARA (as low as reasonably achievable) principle. Every CT protocol should be scrutinized to ensure that the minimum dose is applied to achieve the desired diagnostic image quality. The scanning parameters should be adapted to the patient’s body size and anatomic region. Pediatric patients may require markedly lower radiation dose than adult patients to achieve the same image quality. A common approach to reducing radiation dose is lowering x-ray tube voltage and/or tube current. Decreasing x-ray tube voltage from 120 kVp to 80 kVp can result in a 70% reduction of radiation for a given tube current (14), but 80 kVp is applicable primarily to pediatric CT imaging

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 26

26

Chapter 1

because of its insufficient x-ray penetration for most adult CT imaging. Weight categorization of pediatric patients has been strongly recommended to optimize tube current to levels much lower than those used for adults. Some specific clinical applications such as lung cancer or colonic polyp screening examinations can be performed with a tube current much lower than that used for standard clinical CT studies, resulting in a substantially lower radiation dose. Transaxial cross-sections of the body vary considerably from head to foot, and some body regions deviate substantially from a circular shape. Thus the tube current can be modified on a section-by-section basis to optimize radiation exposure in each area of the body, instead of maintaining a fixed tube current throughout the scan. For example, the cross-section of the chest is elliptical. The x-ray beam is attenuated less when it traverses the chest from an anterior to posterior direction than when it traverses the chest from a lateral direction. This attenuation difference can be exploited to reduce the tube current as the x-beam rotates around the chest, while maintaining a constant signal-to-noise level. This approach has been implemented in modern CT scanners and is widely available as an anatomic tube current modulation technique. Two types of automatic tube current modulation are available in CT scanners today: angular (transverse plane) and longitudinal modulation. The angular modulation technique adjusts the tube current to the patient geometry during each rotation of the scanner to compensate for highly varying x-ray attenuation in asymmetric body regions such as the shoulders and pelvis. The tube current modulation is determined by means of an analysis of anteroposterior and lateral projection topograms or in real-time by evaluating the signal from the detectors. The longitudinal modulation technique adjusts the tube current in the z-direction to lower the dose or maintain adequate dose when moving to different body regions, such as from chest to abdomen. Currently, there are several different implementations of how tube current modulation is determined. Modulation can be based on the attenuation for each body region of a “standard-size” patient, a user-defined noise index value and a range of allowed tube currents, or quality of a reference image. The most recent implementation of automatic tube current modulation has combined angular and longitudinal modulation. Reduction in radiation dose may result in increased image noise and decreased image quality. One approach to improving the image quality of low-dose CT images is by a postprocessing method that applies noise reduction filters (20).

EMERGING COMPUTED TOMOGRAPHY APPLICATIONS Cardiac Computed Tomography MDCT, with its markedly improved temporal and spatial resolution, has provided a new opportunity in cardiac CT

imaging. Previously, electron-beam CT (EBCT) was the method of choice for cardiac CT. Although the current temporal resolution of MDCT is not as high as that of EBCT, MDCT has superior spatial resolution and, unlike EBCT, MDCT can be used for the entire spectrum of routine clinical CT examinations. With scan rotation times of 0.3 to 0.5 second and ECG-triggering or ECG-gating scan techniques, MDCT can readily provide motion-free CT images of the heart and coronary arteries. To generate motion-free images of the cardiac or coronary anatomy, a CT scan of the heart is performed synchronously with an acquired ECG signal. There are two types of ECG synchronization: prospective ECG-triggering and retrospective ECG-gating. In prospective ECG-triggering, the heart is scanned in sequential (step-and-shoot) mode at a predetermined delay after the onset of an R wave. The delay can be either relative (e.g., a percentage of the R-R interval) or absolute (in milliseconds), and either forward (i.e., triggered from new R waves) or reverse (i.e., based on a series of preceding R waves). MDCT can obtain multiple parallel sequential sections that cover several levels of the heart simultaneously. In retrospective ECG-gating, the heart is imaged continuously by a spiral scan while the ECG signal is recorded simultaneously. From the acquired scan data, images are reconstructed from a retrospectively selected ECG cardiac phase, usually the diastolic phase. With retrospective gating, radiation is produced and scan data are acquired continuously throughout the cardiac cycle, even though only a portion of the data are used for the reconstructed images. As a result, the radiation dose of a retrospectively ECG-gated study is higher than that of a prospectively ECG-triggered study. Two common applications of cardiac MDCT are coronary artery calcium scoring and coronary artery angiography. Calcium scoring examinations are performed without intravenous contrast medium and typically by using a prospectively ECG-triggered sequential mode. Because the purpose of a calcium scoring study is to quantify coronary artery calcium, which has high intrinsic tissue contrast relative to noncalcified soft tissue, radiation dose can be reduced without diminishing the diagnostic value of the study (11,35). Most other cardiac CT studies for assessing heart morphology, function, and coronary artery anatomy are performed with an intravenously administered contrast agent using a retrospectively ECG-gated spiral mode. The radiation dose of cardiac MDCT reported in the literature varies greatly depending on the scan parameters used (2). A prospectively ECG-triggered sequential scan results in far less radiation dose than a retrospectively ECGgated spiral CT scan. It should be noted that a high scan overlap (e.g., pitch value of 0.3 to 0.4) is commonly used in retrospectively ECG-gated spiral scans to ensure sufficient contiguous data sampling over the volume without a gap in coverage. The low pitch contributes to the higher radiation dose. For comparison, an ECG-triggered scan is performed without overlap and its pitch may be considered 1. (Note: this is a liberal application of the concept of

5063_Lee_Ch01pp0001-0028 10/13/05 12:01 PM Page 27

Basic Principles of Computed Tomography Physics and Technical Considerations

pitch, which generally is defined and applied only with respect to spiral scans). In prospectively ECG-triggered scans, the radiation doses of cardiac CT are similar in magnitude to those of abdominal and pelvic CT studies. Recently, to maintain the benefits of retrospectively ECG-gated spiral CT but reduce patient dose, ECG-controlled dose modulation techniques have been implemented. Prospectively ECG-triggered, sequential scanning has the benefit of lower radiation dose than retrospectively ECGgated spiral scanning, but it does not provide continuous volume coverage with overlapping sections and is subject to section misregistration artifacts. Furthermore, reconstruction of images in different phases of the cardiac cycle for functional evaluation is not possible. With retrospective ECG-gating, scan data are acquired and available for all phases of the cardiac cycle. In most cases, however, only the diastolic phase scan data are used for image reconstruction. Thus for most studies, a high tube current is required only during diastole, and a low tube current is acceptable during the remainder of the cardiac cycle. Therefore modulation of the tube current on-line with prospective ECG control can reduce radiation exposure while retaining the benefits of retrospectively ECG-gated spiral CT. Because ECG-triggered sequential scanning depends on reliable prediction of the patient’s next R-R interval based on the mean of the preceding R-R intervals, this technique is limited in patients with an arrhythmia.

Positron Emission Tomography–Computed Tomography Positron emission tomography (PET) imaging is widely used for the staging and follow-up of cancer patients. Because of the limited spatial resolution of PET, for most clinical applications, PET imaging findings are interpreted in conjunction with correlative anatomic imaging. Although magnetic resonance imaging and ultrasonography may provide the anatomic imaging correlation, CT is by far the most common anatomic imaging study acquired for oncologic applications. In traditional interpretation of PET and CT images, the images from each study are compared side-to-side and visually co-registered to localize abnormalities. With this approach, optimal interpretation may not be achievable, particularly when a large discrepancy in the time interval and image format between CT and PET imaging does not allow adequate coregistration of the two studies. Software registration techniques for combining CT and PET images work reasonably well for a rigid structure such as the head. Software registration of nonrigid anatomy (e.g., chest, abdomen), however, remains more challenging because of body motion and lack of reliable anatomic landmarks. An alternative approach to software registration is hardware registration by means of an imaging device that is capable of both functional and anatomic imaging in a single mechanical frame. Instead of relying on image features, the registration of anatomic and functional images is conducted

27

via the shared mechanical alignment of the hardware imaging device. This concept of a dual imaging device was tested initially with a combined CT and single-photon emission tomography system in early 1990. In 1998, the first PETCT scanner was introduced. The original systems were fully integrated with shared electronics and detectors. To improve performance, however, the design evolved into a tandem system using previously developed CT and PET scanners combined as independent components. This tandem system has become the standard for scanner design today. The primary purpose of combining CT and PET scanners is to provide precise anatomic localization of potential abnormalities identified on the PET images. The secondary purpose is to use the acquired CT data to calculate the photon attenuation correction for the PET emission data. To explain this purpose, the PET imaging acquisition process is briefly described as follows. PET imaging is based on the detection of coincident photons (energy 511 keV) that are released within the patient’s body. The photons are produced when positrons emitted from radionuclides undergo annihilation with electrons. As a photon travels through the body to reach the detector of a PET scanner, it has some probability of attenuation. This photon attenuation is the most important physical effect that can perturb tracer uptake emission images acquired with PET. The tissues in the body attenuate photons nonuniformly and thus result in an erroneous map of radiotracer activity. Without corrections for photon attenuation, the reconstructed image is degraded, affecting both the visual quality and the quantitative accuracy of the PET data. The correction of photon attenuation requires information about the linear attenuation coefficient for 511 keV photons along each line of response of the PET scanner. In conventional PET scanners, the attenuation information is obtained by performing a transmission scan at PET photon energies with an external source that rotates around the patient. The conventional PET transmission scan requires substantial data acquisition time and is of little use except for photon attenuation. With a PET-CT scanner, attenuation correction of the PET data can be performed with the CT image data, which is feasible because a CT image is a map of linear attenuation coefficients but at x-ray energy. CT scans are an order of magnitude faster (i.e., less than 1 minute) than conventional PET transmission scans (i.e., approximately 20 to 30 minutes), which is beneficial both in the scanner throughput and patient comfort. Because the attenuation values in CT are based on effective x-ray energy of 70 keV (i.e., about one half of kVp), it is necessary to convert the CT image values to generate a 511 keV attenuation map for PET emission data. The conversion process is based on segmenting the CT image into regions corresponding to different tissue types and applying linear scaling appropriate for soft tissue regions and bone regions. This conversion approach tends to overcorrect photon attenuation in regions with metallic implants or in regions enhanced with oral and

5063_Lee_Ch01pp0001-0028 2/17/06 9:39 AM Page 28

28

Chapter 1

intravenous contrast medium, resulting in artifacts in the attenuation-corrected PET image. Artifactually high uptakes related to attenuation-correction are absent in noncorrected images and thus can be readily differentiated from true uptake. Another common artifact in PET imaging is spatial mismatch from respiratory or other patient motion between the PET and CT (or any other transmission) scans. One important consideration regarding using the CT data as the transmission scan is that the radiation dose required for a CT scan is much higher than that for a conventional PET transmission scan. With a standard radiationdose CT scan, the radiation dose for a whole-body PET-CT study might exceed twice the radiation dose for a combined conventional PET emission and transmission scan. To reduce radiation dose, a low radiation-dose CT scan should be obtained when the sole purpose of the CT scan is for photon attenuation correction. In contrast, CT images acquired from the PET-CT scans without oral and intravenous contrast medium, although sufficient for photon attenuation correction and anatomic correlation often are not diagnostically adequate CT examinations. It is, therefore not uncommon to obtain a separate fully diagnostic contrast-enhanced CT scan on a dedicated CT scanner. One approach to reducing radiation dose and at the same time improving the efficiency of scanner utilization is to obtain diagnostically adequate contrast-enhanced CT images during the PET-CT scan. In summary, PET-CT provides both functional and anatomic information in a single imaging study. The CT data are used to correct the PET photon attenuation and to provide anatomic correlation for the PET interpretation. Because PET-CT is a relatively new imaging technique and integrates two different disciplines, the imaging protocols remain controversial and continue to evolve.

REFERENCES 1. Atherton JV, Huda W. Energy imparted and effective doses in computed tomography. Med Phys. 1996;23:735–741. 2. Bae KT, Hong C, Whiting BR. Radiation dose in multidetector row computed tomography cardiac imaging. J Magn Reson Imaging. 2004;19:859–863. 3. Betsou S, Efstathopoulos EP, Katritsis D, et al. Patient radiation doses during cardiac catheterization procedures. Br J Radiol. 1998;71:634–639. 4. Bruder H, Kachelriess M, Schaller S, et al. Single-slice rebinning reconstruction in spiral cone-beam computed tomography. IEEE Trans Med Imaging. 2000;19:873–887. 5. Dainty JC, Shaw R. Image science. New York: Academic Press, 1974:171–183. 6. Feldkamp L, Davis L, Kress, J. Practical conebeam algorithm. J Opt Soc Am. 1984;1:612–619. 7. Flohr T, Stierstorfer K, Bruder H, et al. Image reconstruction and image quality evaluation for a 16-slice CT scanner. Med Phys. 2003;30:832–845. 8. Flohr TG, Schaller S, Stierstorfer K, et al. Multi-detector row CT systems and image-reconstruction techniques. Radiology 2005; 235:756–773.

9. Grass M, Kohler T, Proksa R. 3D cone-beam CT reconstruction for circular trajectories. Phys Med Biol. 2000;45:329–347. 10. Hamberg LM, Rhea JT, Hunter GJ, et al. Multi-detector row CT: radiation dose characteristics. Radiology 2003;226:762–772. 11. Hong C, Bae KT, Pilgram TK. Coronary artery calcium: accuracy and reproducibility of measurements with multi-detector row CT—assessment of effects of different thresholds and quantification methods. Radiology 2003;227:795–801. 12. Hsieh J, Toth TL, Simoni P, et al. A generalized helical reconstruction algorithm for multidetector row CT (abstract). Radiology 2001;221(P):217. 13. Hsieh J. Computed tomography. Bellingham, WA: SPIE, 2003:1–12. 14. Huda W, Scalzetti EM, Levin G. Technique factors and image quality as functions of patient weight at abdominal CT. Radiology 2000;217:430–435. 15. Hunold P, Vogt FM, Schmermund A, et al. Radiation exposure during cardiac CT: effective doses at multi-detector row CT and electron-beam CT. Radiology 2003;226:145–152. 16. Kachelriess M, Schaller S, Kalender WA. Advanced single-slice rebinning in cone-beam spiral CT. Med Phys. 2000;27:754–772. 17. Kachelriess M, Watzke O, Kalender WA. Generalized multidimensional adaptive filtering for conventional and spiral singleslice, multi-slice, and cone-beam CT. Med Phys. 2001;28:475–490. 18. Kak A, Slaney M. Principles of computerized tomography. New York: IEEE Press, 1988:56–92. 19. Kalender WA. Computed tomography. Munich, Germany: Publicis MCD Verlag, 2000:35–55. 20. Kalra MK, Wittram C, Maher MM, et al. Can noise reduction filters improve low-radiation-dose chest CT images? Pilot study. Radiology 2003;228:257–264. 21. Kohler T, Proksa R, Bontus C, et al. Artifact analysis of approximate helical cone-beam CT reconstruction algorithms. Med Phys. 2002;29:51–64. 22. Leung KC, Martin CJ. Effective doses for coronary angiography. Br J Radiol. 1996;69:426–431. 23. Mayo JR, Aldrich J, Muller NL. Radiation exposure at chest CT: a statement of the Fleischner Society. Radiology 2003;228:15–21. 24. McCollough CH, Schueler BA. Calculation of effective dose. Med Phys. 2000;27:828–837. 25. McCollough CH, Zink FE. Performance evaluation of a multislice CT system. Med Phys. 1999;26:2223–2230. 26. Mettler FA Jr, Briggs JE, Carchman R, et al. Use of radiology in U.S. general short-term hospitals: 1980–1990. Radiology 1993; 189:377–380. 27. Morin RL, Gerber TC, McCollough CH. Radiation dose in computed tomography of the heart. Circulation 2003;107:917–922. 28. Nagel HD, ed. Radiation exposure in computed tomography. Hamburg: European Coordination Committee of the Radiological and Electromedical Industries, 2000. 29. Polacin A, Kalendar WA, Brink J, et al. Measurement of slice sensitivity profiles in spiral CT. Med Phys. 1994;21:133–140. 30. Polacin A, Kalendar WA, Marchal G. Evaluation of section sensitivity profiles and image noise in spiral CT. Radiology 1992;185: 29–35. 31. Proksa R, Kohler T, Grass M, et al. The n-PI-method for helical cone-beam CT. IEEE Trans Med Imaging. 2000;19:848–863. 32. Schaller S, Flohr T, Klingenbeck K, et al. Spiral interpolation algorithm for multislice spiral CT—part I: theory. IEEE Trans Med Imaging. 2000;19:822–834. 33. Shrimpton PC, Edyvean S. CT scanner dosimetry. Br J Radiol. 1998;71:1–3. 34. Taguchi K, Aradate H. Algorithm for image reconstruction in multi-slice helical CT. Med Phys. 1998;25:550–561. 35. Takahashi N, Bae KT. Quantification of coronary artery calcium with multi-detector row CT: assessing interscan variability with different tube currents pilot study. Radiology 2003;228: 101–106. 36. Thomton FJ, Paulson EK, Yoshizumi TT, et al. Single versus multi-detector row CT: comparison of radiation doses and dose profiles. Acad Radiol. 2003;10:379–385.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 29

Magnetic Resonance Imaging Principles and Applications Mark A. Brown

2

Richard C. Semelka

PRODUCTION OF NET MAGNETIZATION Magnetic resonance (MR) is based on the interaction between an applied magnetic field and a nucleus that possesses spin. Nuclear spin is one of several intrinsic properties of an atom, and its value depends on the particular atomic composition. Most elements in the periodic table have at least one naturally occurring isotope that possesses spin. Thus, in principle, nearly every element can be examined using MR, and the basic ideas of resonance absorption and relaxation are common for all these elements. The precise details vary from nucleus to nucleus and from system to system. Atoms consist of three fundamental particles: protons, which possess a positive charge; neutrons, which have no charge; and electrons, which have a negative charge. The protons and neutrons are located in the nucleus or core of an atom, whereas the electrons are located in shells, or orbitals, surrounding the nucleus. The characteristic chemical reactions of elements depend on the particular number of each of these particles. Two properties commonly used to categorize elements are the atomic number and the atomic weight. The atomic number is the number of protons in the nucleus and is the primary index used to differentiate atoms. All atoms of an element have the same atomic number. The atomic weight is the sum of the number of protons and the number of neutrons. Atoms with

the same atomic number but different atomic weights are called isotopes. A third property of the nucleus is spin, or intrinsic spin angular momentum. The nucleus can be considered to be continuously rotating about an axis at a constant rate or velocity. This self-rotation axis is perpendicular to the direction of rotation (Fig. 2-1). A limited number of values for the spin are found in nature; that is, the spin, I, is quantized to certain discrete values. These values depend on the atomic number and atomic weight of the particular nucleus. There are two groups of values for I: integral and half-integral values. A nucleus has an integral value for I (e.g., 0, 1, 2, 3) if it has an even atomic weight. If the atomic number is even, then I
0. Such a nucleus does not interact with an external magnetic field and cannot be studied using MR. If the atomic number is odd, then I is nonzero. A nucleus has a half-integral value for I (e.g., 1/2, 3/2, 5/2) if it has an odd atomic weight. Table 2-1 lists the spin and isotopic compositions of some elements commonly found in biologic systems. The 1H nucleus, consisting of a single proton, is a natural choice for probing the body using MR techniques for several reasons. It has a spin of 1/2 and is the most abundant isotope of hydrogen. Its response to an applied magnetic field is one of the largest found in nature. Finally, the body is composed of tissues that contain primarily water and fat, both of which contain hydrogen.

29

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 30

30

Chapter 2

B

N

S

Figure 2-1 A rotating nucleus with a positive charge produces a magnetic field known as the magnetic moment, B, oriented parallel to the axis of rotation (left). This arrangement is analogous to a bar magnet in which the magnetic field is considered to be oriented from the south to the north pole (right).

While a rigorous mathematical description of a nucleus with spin and its interactions requires the use of quantum mechanical principles, most of MR can be described using the concepts of classical mechanics, particularly in describing the actions of a nucleus with spin. The subsequent discussions of MR phenomena in this book use a classical approach. In addition, while the concepts of resonance absorption and relaxation apply to all nuclei with spin, the descriptions in this chapter focus on 1H (commonly referred to as a proton) since most imaging experiments visualize the 1H nucleus. Concurrent with the spinning, positively charged nucleus (the location of the protons) is a local magnetic field known as a magnetic moment. A bar magnet provides a useful analogy. A bar magnet has a north and a south pole or, more precisely, a magnitude and orientation, or direction, to the magnetic field can be defined. A nucleus with spin has an axis of rotation that can be viewed as a vector with a definite orientation and magnitude (see Fig. 2-1). The magnetic moment for the nucleus is parallel to the axis of rotation. This orientation of the nuclear spin and the changes induced in it due to the experimental manipulations that the nucleus undergoes provide the basis for the MR signal. In general, MR measurements are made on collections of similar spins rather than on an individual spin. It is useful to consider such a collection both as individual spins acting independently (a “microscopic” picture) and as a single entity (a “macroscopic” picture). For many concepts, the two pictures provide equivalent results, even though the microscopic picture is more complete. For most concepts presented in this chapter, the macroscopic picture is sufficient for an adequate description. When necessary, the microscopic picture is used.

Consider an arbitrary volume of tissue containing hydrogen atoms (protons). Each proton has a spin vector of equal magnitude. However, the spin vectors for the entire collection of protons within the tissue are randomly oriented in all directions. Performing a vector addition of these spin vectors produces a zero sum; that is, no net magnetization is observed in the tissue (Fig. 2-2). If the tissue is placed inside a magnetic field B0,1 the individual protons begin to rotate perpendicular to, or precess about, the magnetic field. The protons are tilted slightly away from the axis of the magnetic field, but the axis of rotation is parallel to B0. This precession is at a constant rate and occurs because of the interaction of the magnetic field with the spinning positive charge of the nucleus. By convention, B0 and the axis of precession are defined to be oriented in the z direction of a Cartesian coordinate system. The motion of each proton can be described by a unique set of coordinates perpendicular (x and y) and parallel (z) to B0. The perpendicular, or transverse, coordinates are nonzero and vary with time as the proton precesses, but the z coordinate is constant with time (Fig. 2-3). The rate, or frequency, of precession is proportional to the strength of the magnetic field and is expressed by Eq. 1, the Larmor equation: ω0
γB0>2π

(1)

where ω 0 is the Larmor frequency in megahertz (MHz),2 B0 is the magnetic field strength in Tesla (T) that the proton experiences, and γ is a constant for each nucleus in s–1T–1, known as the gyromagnetic ratio. Values for ω 0 and γ at 1.5 T for several nuclei are tabulated in Table 2-1. If a vector addition is performed, as before, for the spin vectors inside the magnetic field, the results will be slightly different from those for the sum outside the field. In the direction perpendicular to B0, the spin orientations are still randomly distributed just as they were outside the magnetic field, in spite of the time-varying nature of each transverse component. There is still no net magnetization perpendicular to B0. However, in the direction parallel to the magnetic field, there is a different result. Because there is an orientation to the precessional axis of the proton that is constant with time, there is a constant, nonzero interaction, or coupling, between the proton and B0. This coupling causes a difference in energy between

1 Vector quantities with direction and magnitude are indicated by boldface type, while scalar quantities that are magnitude only are indicated by regular typeface. 2 The factor of 2 in Eq. 1 is necessary to convert from angular to cyclical frequency. In many discussions of MR,  (Greek letter omega) is used to represent angular frequency, with units of s–1, while cyclical frequency, in units of Hz, is represented by  (Greek letter nu) or f. The Larmor equation is more properly written as 0
 B0/2. In imaging presentations, the Larmor equation is expressed as Eq. 1, using  but with units of Hertz. To minimize confusion, we follow the imaging tradition throughout this chapter.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 31

Magnetic Resonance Imaging Principles and Applications

31

TABLE 2-1 SELECTED ELEMENTS OF THE PERIODIC TABLE Nuclear composition Element 1H,

Protium Deuterium 3He 6Li 7Li 12C 13C 16O 17O 19F 23Na 31P 2H,

Protons

Neutrons

Nuclear Spin, I

Gyromagnetic ratio (MHz T1)

1 1 2 3 3 6 6 8 8 9 11 15

0 1 1 3 4 6 7 8 9 10 12 16

1/2 1 1/2 1 3/2 0 1/2 0 5/2 1/2 3/2 1/2

42.5774 6.53896 32.436 6.26613 16.5483 0 10.7084 0 5.7743 40.0776 11.2686 17.2514


at 1.5 T

% Natural abundance

(MHz)

99.985 0.015 0.000138 7.5 92.5 98.90 1.10 99.762 0.038 100 100 100

63.8646 9.8036 48.6540 9.39919 24.8224 0 16.0621 0 8.6614 60.1164 16.9029 25.8771

Adapted from Mills I, ed. Quantities, units, and symbols in physical chemistry. International Union of Pure and Applied Chemistry, Physical Chemistry Division. Oxford, UK: Blackwell Science, 1989.

protons aligned parallel to or along B0 and those aligned antiparallel to or against B0. This energy difference, DE, is proportional to B0 (Fig. 2-4). The orientation that is parallel to B0 (also known as spin up) is of lower energy than the antiparallel orientation (also known as spin down). For a collection of protons, more will be oriented parallel to B0 than antiparallel to B0; that is, there is an induced polarization of the spin orientation by the magnetic field (Fig. 2-5A). The distribution of protons in each energy level may be calculated using the Boltzmann distribution function. The exact number of protons in each level depends on B0 (through DE) and increases with increasing B0. For a collection of protons at body temperature (310 K) at 1.5 T, there will typically be an excess of
1:106 protons

in the lower level out of the approximately 1025 protons within the tissue volume. This unequal number of protons in each energy level means that the vector sum of spins will be nonzero and will point parallel to the magnetic field. In other words, the tissue will become polarized or magnetized in the presence of B0 with a value M0, known as the net magnetization. The orientation of this

z 0

B0

y z component y component x component

x

0 Figure 2-3 Inside a magnetic field, a proton precesses or reFigure 2-2 Microscopic and macroscopic pictures of a collection of protons in the absence of an external magnetic field. In the absence of a field, the protons have their spin vectors oriented randomly (microscopic picture, left). The vector sum of these spin vectors is zero (macroscopic picture, right).

volves about the magnetic field. The precessional axis is parallel to the main magnetic field, B0. The z component of the spin vector (projection of the spin onto the z axis) is the component of interest because it does not change in magnitude or direction as the proton precesses. The x and y components vary with time at a frequency ω0 proportional to B0 as expressed by Eq. 1.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 32

32

Chapter 2

Higher Energy

∆E

B0

Lower Energy

Figure 2-4 Zeeman diagram. In the absence of a magnetic field (left), a collection of protons will be equal in energy so that there is no preferential alignment. In the presence of a magnetic field (right), the spin up orientation (z component parallel to the magnetic field, B0) is of lower energy and its configuration contains more protons than the higher energy, spin down configuration. The difference in energy, ∆E, between the two levels is proportional to B0.

net magnetization will be in the same direction as B0 and will be constant with respect to time (Fig. 2-5B). For tissues in the body, the magnitude of M0 is proportional to B0 (equation 2): M0
χ B0

(2)

where χ is known as the bulk magnetic susceptibility or simply the magnetic susceptibility. This arrangement, with M0 aligned along the magnetic field with no transverse component, is the normal, or equilibrium, configuration for the protons. This configuration of spins has the lowest energy and is the arrangement to which the protons will naturally try to return following any perturbations, such as energy absorption. This induced magnetization, M0, is the source of signal for all MR experiments. Consequently, all other things being equal, the greater the field strength, the greater the value of M0 and the greater the potential MR signal. z

B0

z

B0

y

y

M0 x

A

x

B

Figure 2-5 Microscopic (A) and macroscopic (B) pictures of a collection of protons in the presence of an external magnetic field. Each proton precesses about the magnetic field, B0. If a rotating frame of reference is used with a rotation rate of  0, then the collection of protons appears stationary. While the z components are one of two values (one positive and one negative), the x and y components can be any value, positive or negative. The protons will appear to track along two “cones,” one with a positive z component and one with a negative z component. Because there are more protons in the upper cone, there will be a nonzero vector sum, M0, the net magnetization. It will be of constant magnitude and parallel to B0.

Another way to visualize this net magnetization is to recall that the individual spins precess about the magnetic field. When the spins absorb energy, this precessional motion continues but becomes more complicated to describe. A useful technique to simplify the description is called a rotating frame of reference, or rotating coordinate system. In a rotating frame, the coordinate system rotates about one axis while the other two axes vary with time. By choosing a suitable axis and rate of rotation for the coordinate system, the moving object appears stationary. For MR experiments, a convenient rotating frame uses the z axis, parallel to B0, as the axis of rotation while the x and y axes rotate at the Larmor frequency, ω0. When viewed in this fashion, the precessing spin appears stationary in space with a fixed set of x, y, and z coordinates. If the entire collection of protons in the tissue volume is examined, a complete range of x and y values will be found, both positive and negative, but only two z values. There will be an equal number of positive and negative x and y values, but a slight excess of positive z values, as just described. If a vector sum is performed on this collection of protons, the x and y components sum to zero but a nonzero, positive z component will be left, the net magnetization M0. In addition, since the z axis is the axis of rotation, M0 does not vary with time. Regardless of whether a stationary or fixed coordinate system is used, M0 is of fixed amplitude and is parallel to the main magnetic field. For all subsequent discussions in this book, a rotating frame of reference with the rotation axis parallel to B0 is used when describing the motion of the protons.

CONCEPTS OF MAGNETIC RESONANCE The MR experiment, in its most basic form, can be analyzed in terms of energy transfer. During the measurement, the patient, or sample, is exposed to energy at the correct frequency that will be absorbed. A short time later, this energy is reemitted, at which time it can be detected and processed. A detailed presentation of the processes involved in this absorption and reemission are beyond the scope of this text. However, a general description of the nature of the molecular interactions is useful. In particular, the relationship between the molecular picture and the macroscopic picture provides an avenue for the explanation of the principles of MR. The formation of the net magnetization, M0, by the protons within a sample was described in the previous section. The entire field of MR is based on the manipulation of M0. The simplest manipulation involves the application of a short burst, or pulse, of radiofrequency (rf ) energy. This pulse, also known as an excitation pulse, contains many frequencies spread over a narrow range or bandwidth. During the pulse, the protons absorb the portion of this energy at a particular frequency. Following the pulse, the protons reemit the energy at the same frequency.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 33

33

Magnetic Resonance Imaging Principles and Applications z

z y

B0 B0

∆E

y 90-degree pulse

M B1

x

M x

Figure 2-7 Energy absorption (macroscopic). In a rotating frame

Figure 2-6 Energy absorption (microscopic). The difference in

energy ∆E between the two configurations (spin up and spin down) is proportional to the magnetic field strength B0 and the corresponding precessional frequency ω 0. When energy at this frequency is applied, a spin from the lower energy state is excited to the upper energy state. Also, a spin from the upper energy state is stimulated to give up its energy and relax to the lower energy state. Because there are more spins in the lower energy state, there is a net absorption of energy by the spins in the sample.

The particular frequency absorbed is proportional to the magnetic field B0; the equation relating the two is the Larmor equation, Eq. 1. The frequency of energy absorbed by an individual proton is defined very precisely by the magnetic field that the proton experiences due to the quantized nature of the spin values. When a proton is irradiated with energy of the correct frequency (ω 0 ), it will be excited from the lower energy (spin up) orientation to the higher energy (spin down) orientation (Fig. 2-6). At the same time, a proton in the higher energy level will be stimulated to release its energy and will go to the lower energy level. The energy difference (∆E) between the two levels is exactly proportional to the frequency ω 0 and thus the magnetic field B0. Only energy at this frequency stimulates transitions between the spin up and spin down energy levels. This quantized energy absorption is known as resonance absorption and the frequency of energy is known as the resonant frequency. Although an individual proton absorbs the rf energy, it is more useful to discuss the resonance condition by examining the effect of the energy absorption on the net magnetization, M0. The energy is applied as an rf pulse with a central frequency ω 0 and an orientation perpendicular to B0, as indicated by an effective field B1 (Fig. 2-7). This orientation difference allows a coupling between the rf pulse and M0 so that energy can be transferred to the protons. Absorption of the rf energy of frequency ω 0 causes M0 to rotate away from its equilibrium orientation. The direction of rotation of M0 is perpendicular to both B0 and B1. If the transmitter is left on long enough and at a high enough amplitude, the absorbed energy causes M0 to rotate entirely into the transverse plane, a result known as a 90-degree pulse. When viewed in a rotating frame, the motion of M0 is a simple vector rotation; however, the end result is the same whether a rotating or stationary frame of reference is used.

of reference, the rf pulse broadcast at the resonant frequency ω 0 can be treated as an additional magnetic field, B1, oriented perpendicular to the main magnetic field, B0. When energy is applied at the appropriate frequency, the protons absorb it and the net magnetization, M, rotates into the transverse plane. The direction of rotation is perpendicular to both B0 and B1. The amount of resulting rotation of M is known as the pulse flip angle.

When the transmitter is turned off, the protons immediately begin to realign themselves and return to their original equilibrium orientation. They emit energy at frequency ω 0 as they do so through a process known as relaxation. In addition, M0 begins to precess about B0, similar to the behavior of a gyroscope when tilted away from the vertical axis. If a loop of wire (receiver) is placed perpendicular to the transverse plane, the protons induce a voltage in the wire during their precession. This induced voltage, the MR signal, is known as the free induction decay (FID) (Fig. 2-8A). The initial magnitude of the FID signal depends on the value of M0 immediately prior to the 90-degree pulse. The FID decays with time as more of the protons give up their absorbed energy through a process known as relaxation (see below), and the coherence, or uniformity, of the proton motion is lost. In general, three aspects of an MR signal are of interest: its magnitude, or amplitude, its frequency, and its phase relative to the rf transmitter phase (Fig. 2-9). As mentioned previously, the signal magnitude is related to the value of M0 immediately prior to the rf pulse. The signal frequency is related to the magnetic field influencing the protons. If all the protons were to experience the same magnetic field, B0, then only one frequency would be present within the FID. In reality, there are many magnetic fields throughout the magnet, and thus, many MR signals at many frequencies following the rf pulse. These signals are superimposed so that the FID contains many frequencies varying as a function of time. It is easier to analyze such a multicomponent signal in terms of frequency rather than of time. The conversion of the signal amplitudes from a function of time to a function of frequency is accomplished using a mathematical operation called the Fourier transformation. In the frequency presentation or frequency domain spectrum, the MR signal is mapped according to its frequency relative to a reference frequency, typically the transmitter frequency ω TR. For systems using quadrature detectors (see section on hardware), ω TR is centered in the display with frequencies higher and lower than ω TR located to the left

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 34

34

Chapter 2

MAGNITUDE REAL

ω TR + ω NQ 0

0 Time

A

PHASE

IMAGINARY

ω TR Frequency

Time

ω TR - ω NQ ωTR + ωNQ

ωTR

ω TR - ω NQ

Frequency

Figure 2-8 Free induction decay, real and imaginary. A: The response of the net magnetization, M, to an rf pulse as a function of time is known as the free induction decay (FID). It is proportional to the amount of transverse magnetization generated by the pulse. The FID is maximized when using a 90-degree excitation pulse. B: Fourier transformation of (A), magnitude and phase. The Fourier transformation is used to convert the digital version of the MR signal (FID) from a function of time to a function of frequency. Signals measured with a quadrature detector are displayed with the transmitter (reference) frequency ω TR in the middle of the display. The Nyquist frequency ω NQ below and above ω TR are the minimum and maximum frequencies of the frequency display, respectively. For historical reasons, frequencies are plotted with lower frequencies on the right side and higher frequencies on the left side of the display.

and right, respectively (Fig. 2-8B). The frequency domain thus allows a simple way to examine the magnetic environment that a proton experiences. This simplification is not without a penalty. Use of the Fourier transformation causes loss of the ability to directly relate signal intensities to the number of protons producing the signal. However, the signal intensity can be related from one frequency to another within the same measurement; that is, only relative signal intensities can be compared. Since the proton precession is continuous, the MR signal is continuous, or analog, in nature. However, postprocessing techniques, such as Fourier transformation, require a digital representation of the signal. To produce a digital version, the FID signal is measured, or sampled, using an b

b a

a c

c

Figure 2-9 Planar (left) and circular (right) representations of a time-varying wave. The amplitude (a) is the maximum deviation of the wave from its mean value. The period (b) is the time required for completion of one complete cycle of the wave. The frequency of the wave is the reciprocal of the period. The phase or phase angle of the wave (c) describes the shift in the wave relative to a reference (a second wave for the planar representation, horizontal axis for circular representation). The two plane waves displayed have the same amplitude and period (frequency), but have a phase difference of π/4 or 90 degrees.

analog-to-digital converter (ADC). In most instances, the resonant frequencies of protons are greater than many ADCs can process. For this reason, a phase-coherent difference signal is generated based on the frequency and phase of the input rf pulse; that is, the signal actually digitized is the measured signal relative to ω TR. Under normal conditions, this so-called demodulated signal is digitized for a predetermined time, known as the sampling time, and with a user-selectable number of data points. In such a situation, there will be a maximum frequency, known as the Nyquist frequency ω NQ, that can be accurately measured: ωNQ
(Total number of data points)/2  (Sampling time) (3) In MR, the Nyquist frequency can be 500 to 500,000 Hz, depending on the combination of sampling time and number of data points. To exclude frequencies greater than the Nyquist limit from the signal, a filter known as a lowpass filter is used prior to digitization. Frequencies excluded by the low-pass filter are usually noise, so that filtering provides a method for improving the signal-to-noise ratio (S/N) for the measurement. The optimum S/N is usually obtained by increasing the sampling time to match the Nyquist frequency and low-pass filter width for the particular measurement conditions. For quadrature detection systems typically used in MR, the total receiver bandwidth is 2  ω NQ centered about ω TR (Fig. 2-8B). The specific frequency that a proton absorbs is dependent on magnetic fields arising from two sources. One is the applied magnetic field, B0. The other one is molecular in origin and produces the chemical shift. In patients, the bulk of the hydrogen MR signal arises from two sources, water and fat. Water has two hydrogen atoms bonded to

B

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 35

Magnetic Resonance Imaging Principles and Applications

one oxygen atom while fat has many hydrogen atoms bonded to a long chain carbon framework (typically 10 to 18 carbon atoms in length). Because of its different molecular environment, a hydrogen atom in fat has a different local magnetic field than one in water. Different resonant frequencies are measured from fat and water protons under the influence of the same main magnetic field. A convenient scale to express these frequency differences is the parts per million (ppm) scale, which is the resonant frequency of the hydrogen atom of interest relative to a reference frequency. Frequency differences expressed in this form are known as chemical shifts. While the choice of reference frequency is arbitrary, it is convenient to use ω TR as the reference. The primary advantage of the ppm scale is that frequency differences are independent of B0. For fat and water, the difference in chemical shifts at all field strengths is approximately 3.5 ppm, with fat at a lower frequency. At 1.5 T, this difference is 220 Hz, whereas at 3.0 T, it is 450 Hz (Fig. 2-10). The chemical shift difference between fat and water can be visualized in the rotating frame. A 150-Hz difference in frequency means that the fat resonance precesses slower than the water resonance by 6.7 ms per cycle (1/150 Hz). The fat resonance will align with or be in phase with the water resonance every 6.7 ms at 1.0 T. For a 1.5-T MR system, the same cycling occurs every 4.5 ms (1/220 Hz), and it occurs every 2.25 ms for a 3.0-T system (Fig. 2-11). The 3.5-ppm chemical shift difference mentioned previously is an approximate difference. The fat resonance signal is a composite from all the protons within the fat molecule. The particular chemical composition (e.g., saturated versus unsaturated hydrocarbon chain, length of hydrocarbon chain) determines the exact resonant frequency for this composite signal. The 3.5-ppm difference applies to the majority of fatty tissues found in the body. Chemical shift differences between protons in different molecular environments provide the basis for MR spectroscopy, which is described in more detail in the section on MR spectroscopy. 1.5 T

3.0 T

225Hz

450Hz

3.5ppm

3.5ppm

Figure 2-10 Spectrum of water and fat at 1.5 T (left) and 3.0 T (right). The resonant frequencies for water and fat are separated by approximately 3.5 ppm, which corresponds to an absolute frequency difference of 225 Hz for a 1.5-T magnetic field (63 MHz) and 450 Hz for a magnetic field of 3.0 T (126 MHz).

35

RELAXATION As mentioned previously, the MR measurement can be analyzed in terms of energy transfer. Relaxation is the process by which the spins release the energy that they absorbed from the rf pulse. Relaxation is a fundamental process in MR, as essential as energy absorption, and provides the primary mechanism for image contrast, as discussed later. In resonance absorption, rf energy is absorbed by the spins only when it is broadcast at the correct frequency. The additional energy disturbs the equilibrium arrangement of spins parallel and antiparallel to B. Following excitation, relaxation occurs, in which the spins release this added energy and return to their original configuration through naturally occurring processes. Although an individual spin absorbs the energy, relaxation times are measured for an entire sample and are statistical or average measurements; that is, relaxation times are measured for liver or spleen as bulk samples rather than for the individual water or fat molecules within the organs. Two relaxation times can be measured, known as T1 and T2. While both times measure the spontaneous energy transfer by an excited proton, they differ in the final disposition of the energy.

T1 Relaxation and Saturation The relaxation time T1 is the time required for the z component of M to return to 63% of its original value following an excitation pulse. It is also known as the spin-lattice relaxation time or longitudinal relaxation time. As mentioned in the previous section, M0 is parallel to B0 at equilibrium, and energy absorption rotates M0 into the transverse plane. T1 relaxation provides the means by which the protons give up their energy to return to their original orientation. If a 90-degree pulse is applied to a sample, M0 will rotate as illustrated in Figure 2-7, and there will be no longitudinal magnetization present following the pulse. As time goes on, a return of the longitudinal magnetization will be observed as the protons release their energy (Fig. 2-12). This return of magnetization follows an exponential growth process, with T1 being the time constant describing the rate of growth. After three T1 time periods, M will have returned to 95% of its value prior to the excitation pulse, M0. The term “spin-lattice” refers to the fact that the excited proton (spin) transfers its energy to its surroundings (lattice) rather than to another spin. The energy no longer contributes to spin excitation. This energy transfer to the surroundings has some very important consequences. Suppose the rf energy is continuously applied at the resonant frequency so that no relaxation occurs. A comparison of the microscopic and macroscopic pictures is useful at this point. In the microscopic picture, the protons in the lower energy level absorb the rf energy, and the protons in the upper energy level are stimulated to emit their energy. Because energy is continuously transmitted, the proton populations of the

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 36

36

Chapter 2

A

B

C Figure 2-11 Precession of fat

D

E

Water

Fat

two levels gradually equalize. When this equalization occurs, no further net absorption of energy is possible, a condition known as saturation. In the macroscopic picture, M rotates continuously but gradually gets smaller in magnitude until it disappears as the net population difference approaches zero. Because there is no net magnetization, there is no coherence of proton motion in the transverse plane and thus no signal is produced. A collection of

1.0 0.95 0.86 0.63

Mz M0

0

T1

2*T1

3*T1

TR Figure 2-12 T1 relaxation curve. Following a 90-degree rf pulse, there is no longitudinal magnetization. A short time later, longitudinal magnetization will be observed as the protons release their energy through T1 relaxation. Gradually, as more protons release their energy, a larger fraction of the z component of the net magnetization, MZ, is reestablished. Eventually, the initial net magnetization, M0, will be restored completely. The change in MZ/M0 with time TR follows an exponential growth process. The time constant for this process is T1, the spin-lattice relaxation time, and is the time when MZ has returned to 63% of its original value.

and water protons. Because of the 3.5-ppm frequency difference, a fat proton precesses at a slower frequency than a water proton. In a rotating frame at the water resonance frequency, the fat proton cycles in and out of phase with the water proton. Following the excitation pulse, the two protons are in phase (A). After a short time, they will be 180 degrees out of phase (B), then in phase (C), then 180 degrees out of phase (D), then in phase (E). The contribution of fat to the total signal fluctuates and depends on when the signal is detected. At 1.5 T, the in-phase times are 0 (A), 4.5 (C), and 9 (E) ms, and the out-of-phase times are 2.25 (B) and 6.7 (D) ms. At 3.0 T, the in-phase times are 0, 2.25, and 4.5 ms, and the out-of-phase times are 1.12 and 3.4 ms, respectively.

protons can absorb only a limited amount of energy before they become saturated. In a modern MR experiment, pulsed rf energy is applied to the protons repeatedly with a delay time between the pulses. This time between pulses allows the excited protons to give up the absorbed energy (T1 relaxation). As the protons give up this energy to their surroundings, the population difference (spin up versus spin down) is reestablished so that net absorption can reoccur after the next pulse. In the macroscopic picture, M returns toward its initial value M0 as more energy is dissipated. Since M is the ultimate source of the MR signal, the more energy is dissipated, the more signal is generated following the next rf pulse. For practical reasons, the time between successive rf pulses is usually insufficient for complete T1 relaxation so that M will not be completely restored to M0. Application of a second rf pulse prior to complete relaxation will rotate M into the transverse plane, but with a smaller magnitude than following the first rf pulse. The following experiment describes the situation (Fig. 2-13): (a) A 90-degree rf pulse is applied. M is rotated into the transverse plane. (b) A repetition time, TR, elapses, insufficient for complete T1 relaxation. The longitudinal magnetization at the end of TR, M’, is less than in (a). (c) A second 90-degree rf pulse is applied. M’ is rotated into the transverse plane. (d) After a second time TR elapses, M” is produced. It is smaller in magnitude than M’, but the difference is less than the difference between M and M’.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 37

Magnetic Resonance Imaging Principles and Applications

a b 90-degree pulse

M

90-degree pulse

time TR

M'

time TR

M"

Figure 2-13 Following a 90-degree rf pulse, longitudinal magnetization is regenerated through T1 relaxation. If the time between successive rf pulses, TR, is insufficient for complete recovery of the net magnetization, M, then only M’ will be present at the time of the next rf pulse (a). If time TR elapses again, then only M’’ will be present (b). M’’ will be smaller than M’, but the difference will be less than the difference between M and M’.

Following a few repetitions, M returns to the same magnitude prior to each rf pulse; that is, M achieves a steadystate value. In general, this steady-state value depends on five parameters: 1. The main magnetic field B0. The larger the value for B0, the larger M. 2. The number of protons producing M (per unit volume of tissue, known as the proton density); 3. The amount of energy absorbed by the protons (the pulse angle or flip angle). 4. The rate of rf pulse application (TR). 5. The efficiency of the protons in releasing their energy (T1 relaxation time). For many MRI experiments, such as standard spin echo and gradient echo imaging, a steady state of M is present because multiple rf pulses are applied and the repetition time, TR, between the pulses is nearly always less than sufficient for complete relaxation. To produce this steady state prior to data collection, additional rf pulses are applied to the tissue immediately prior to the main imaging pulses. These extra rf pulses are known as preparatory pulses or dummy pulses because the generated signals are usually ignored. These preparatory pulses ensure that M has the same magnitude prior to every measurement during the scan. As mentioned earlier, spin-lattice relaxation measures energy transfer from an excited proton to its surroundings. The key to this energy transfer is the presence of some type of molecular motion (e.g., vibration, rotation) in the vicinity of the excited proton with an intrinsic frequency ωL that matches the resonant frequency, ω 0. The closer ω 0 is to ω L, the more readily the motion absorbs the energy and the more frequently this energy transfer occurs, allowing the collection of protons to return to its equilibrium configuration sooner. In tissues, the nature of the protein molecular structure and any metal ions that may be present have a pronounced effect on the particular ω L. Metals ions, such as iron or manganese, can have significant magnetic moments that may influence the local environment. While

37

the particular protein structures are different for many tissues, the molecular rotation, or tumbling, of most proteins typically has an ω L of approximately 1 MHz. Therefore, at higher resonant frequencies (higher B0), there is a poorer match between ω L and ω 0. This enables a less efficient energy transfer to occur, and thus T1 is longer. This is the basis for the frequency dependence of T1; namely, that T1 increases with increasing magnetic field strength. This is also the reason that a larger B0 does not necessarily translate to a greater signal, as saturation is more prevalent due to the longer T1 times.

T2 Relaxation, T2* Relaxation, and Spin Echoes The relaxation time T2 is the time required for the transverse component of M to decay to 37% of its initial value via irreversible processes. It is also known as the spin–spin relaxation time or transverse relaxation time. As mentioned earlier, M0 is oriented only along the z (B0) axis at equilibrium and no portion of M0 is in the xy plane. The coherence is entirely longitudinal. Absorption of energy from a 90-degree rf pulse, as in Figure 2-7, causes M0 to rotate entirely into the xy plane, so that the coherence is in the transverse plane at the end of the pulse. As time elapses, this coherence disappears while, at the same time, the protons release their energy and reorient themselves along B0. This disappearing coherence produces the FID described previously. As this coherence disappears, the value of M in the xy plane decreases toward zero. T2, or T2*, relaxation is the process by which this transverse magnetization is lost. A comparison of the microscopic and macroscopic pictures provides additional insight. At the end of the 90-degree rf pulse, when the protons have absorbed energy and are oriented in the transverse plane, each proton precesses at the same frequency, ω 0, and is synchronized at the same point or phase of its precessional cycle. Since a nearby proton of the same type has the same molecular environment and the same ω 0, it can readily absorb the energy that is being released by its neighbor. Spin–spin relaxation refers to this energy transfer from an excited proton to another nearby proton (Fig. 2-14). The absorbed energy remains as spin excitation rather than being transferred to the surroundings, as in T1 relaxation. This proton–proton energy transfer can occur many times as long as the protons are in close proximity and remain at the same ω 0. Intermolecular and intramolecular interactions, such as vibrations and rotations, cause ω 0 to fluctuate. This fluctuation produces a gradual, irreversible loss of phase coherence to the spins as they exchange the energy and reduce the magnitude of the transverse magnetization and the generated signal (Fig. 215). T2 is the time when the transverse magnetization is 37% of its value immediately after the 90-degree pulse, when this irreversible process is the only cause for the loss of coherence. As more time elapses, this transverse

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 38

38

Chapter 2

Figure 2-14 Spin–spin relaxation. A: Two water molecules. One spin on one molecule has absorbed rf energy and is excited (spin down). B: If the spins are in close proximity, the energy can be transferred from the first molecule to a spin on the second molecule.

coherence completely disappears, only to reform in the longitudinal direction as T1 relaxation occurs. This dephasing time, T2, is always less than or equal to T1. There are several potential causes for a loss of transverse coherence to M. One is the movement of the adjacent spins due to molecular vibrations or rotations. This movement is responsible for spin–spin relaxation, or the true T2. Another cause arises from the fact that a proton never experiences a magnetic field that is 100% uniform or homogeneous. As the proton precesses, it experiences a fluctuating local magnetic field, causing a change in ω 0 and a loss in transverse phase coherence. This nonuniformity in B0 comes from three sources: Main field inhomogeneity. There is always some degree of nonuniformity to B0 due to imperfections in magnet manufacturing, composition of nearby building walls, or other sources of metal. This field distortion is constant during the measurement time. Sample-induced inhomogeneity. Differences in the magnetic susceptibility or degree of magnetic polarization of adjacent tissues (e.g., bone, air) distort the local magnetic field near the interface between the different tissues. This

1i

ii2

3 iii

v5

6 vi

A

H

O

H

H

H

H

H

O

B

H

O

H

O

90 o pulse

4iv

A 2 1.0

MXY MXY

3

max

0.37 4 5 6 0

B

T2

TE

Figure 2-15 A: A rotating frame slower than ω 0 is assumed for this figure. The net magnetization, M (arrow), is oriented parallel to B0 (not shown) prior to the pulse (1). Following a 90-degree rf pulse, the protons initially precess in phase in the transverse plane (2). Due to inter- and intramolecular interactions, the protons begin to precess at different frequencies (dashed arrow, faster; dotted arrow, slower) and become asynchronous with each other (3). As more time elapses (4, 5), the transverse coherence becomes smaller until there is complete randomness of the transverse components and no coherence (6). B: Plot of the relative xy component of the net magnetization, MXY, as a function of the echo time. The numbers correspond to the expected MXY component from Figure 2-15A. The change in MXY/MXYmax with time follows an exponential decay process. The time constant for this process is the spin–spin relaxation time T2 and is the time when MXY has decayed to 37% of its original value.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 39

Magnetic Resonance Imaging Principles and Applications

inhomogeneity is of constant magnitude and is present as long as the patient is present within the magnet. Imaging gradients. As discussed in the next section, the technique used for spatial localization generates a magnetic field inhomogeneity that induces proton dephasing. This inhomogeneity is transient during the measurement. Proper design of the pulse sequence eliminates the imaging gradients as a source of dephasing. The other sources that contribute to the total transverse relaxation time, T2*, include the dephasing due to the main field inhomogeneity and magnetic susceptibility differences. The decay of the transverse magnetization following a 90-degree rf pulse, the FID, follows an exponential process with a time constant of T2* rather than T2. For most tissues or liquids, main field inhomogeneity is the major factor determining T2*, whereas for tissue with significant iron deposits or air-filled cavities, magnetic susceptibility differences predominate T2*. Some sources of proton dephasing can be reversed by the application of a 180-degree rf pulse, which is described by the following sequence of events (Fig. 2-16): 1. 2. 3. 4.

A 90-degree rf pulse; A short delay of time, t; A 180-degree rf pulse; and A second time delay, t.

A

The initial 90-degree rf pulse rotates M0 into the transverse plane. During the time t, proton dephasing occurs through T2* relaxation processes and the transverse coherence diminishes. Application of the 180-degree rf pulse causes the protons to reverse their phases relative to the resonant frequency. The rates and directions of precession for the protons do not change, only their relative phase. If time t elapses again, then the protons will regain their transverse coherence. This reformation of phase coherence induces another signal in the receiver coil, known as a spin echo. Sources of dephasing that do not change during the two time periods, the main field inhomogeneity and magnetic susceptibility differences, are eliminated because the protons experience exactly the same interactions prior to and following the 180-degree pulse. This means that the contributions to T2* relaxation from these static sources disappear. Only the irreversible spin–spin relaxation is unaffected by the 180-degree rf pulse, so that the loss of phase coherence and signal amplitude for a spin echo is due only to true T2 relaxation. Following echo formation, the protons continue to precess and dephase a second time as the sources of dephasing continue to affect them. Application of a second 180-degree rf pulse again reverses the proton phases and generates another coherence to the protons, producing another spin echo. This second echo differs from the first echo by the

C

B

time TE /2

90-degree pulse

E

D

180-degree pulse

39

time TE /2

Figure 2-16 A rotation frame slower than ω 0 is assumed for this figure. The net magnetization, M (arrow), is oriented parallel to B0 (not shown) prior to the pulse (A). Application of a 90-degree rf pulse rotates M into the transverse plane (B). Due to T2* relaxation processes, the protons become asynchronous with each other during the echo time (TE)/2 (C). Application of a 180-degree rf pulse causes the protons to reverse their phase relative to the transmitter phase. The protons that precessed most rapidly are farthest behind (dashed arrow), while the slowest protons are in front (dotted arrow) (D). Allowing time TE/2 to elapse again allows the protons to regain their phase coherence in the transverse plane (E), generating a signal in the receiver coil known as a spin echo. The loss in magnitude of the reformed coherence relative to the original coherence (B) is due to irreversible processes (i.e., true spin–spin or T2 relaxation).

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 40

40

Chapter 2

greater amount of T2 relaxation contributing to the signal loss. This process of spin echo formation by 180-degree rf pulses can be repeated as many times as desired, until T2 relaxation completely dephases the protons. The use of multiple 180-degree pulses maintains phase coherence to the protons longer than the use of a single 180-degree rf pulse because of the significant dephasing that the field inhomogeneity induces over very short time periods. One important difference between T1 and T2 relaxation is in the influence of B0. As mentioned earlier, T1 is very sensitive to B0, with longer T1 times measured for a tissue at higher values of B0. T2 is relatively insensitive to B0 at the relatively large field strengths currently used in MRI. Only at very low values of B0 (less than 0.05 T) will there be significant changes in T2. The other components of T2*, main-field inhomogeneity and magnetic susceptibility differences, become more prominent at higher values of B0. Good magnetic field uniformity is more difficult to generate at high magnetic fields. Larger values of B0 also cause greater differences in M0 between two tissues with different magnetic susceptibilities. The result is that T2weighted techniques show little sensitivity to B0, while T2*-weighted techniques show greater signal differences at higher values of B0.

PRINCIPLES OF MAGNETIC RESONANCE IMAGING The relationship between the frequency of energy that a proton absorbs and the magnetic field strength that it experiences was described previously. MRI uses this field dependence to localize these proton frequencies to different regions of space. In MRI, the magnetic field is made spatially dependent through the application of magnetic field gradients. These gradients are small perturbations superimposed on the main magnetic field B0, with a typical imaging gradient producing a total field variation of less than 1%. They are also linear perturbations to B0, so that the exact magnetic field is linearly dependent on the location inside the magnet. Gradients are also applied for short periods of time during a scan and are referred to as gradient pulses. In clinical MRI, the magnetic field gradients produce linear variations primarily in one direction only, one in each of the x, y, and z directions, known as the physical gradients. Each gradient is assigned, through the operating software, to one or more of the three “logical,” or functional, gradients required to obtain an image: slice selection, readout (or frequency encoding), and phase encoding. The particular pairing of physical and logical gradients is variable and depends on the acquisition parameters and patient positioning as well as the particular manufacturer’s choice of physical directions. The combination of gradient pulses, rf pulses, data sampling periods, and the timing between them that is used to acquire an image is known as a pulse sequence.

The presence of magnetic field gradients requires an expanded version of the Larmor equation given in Eq. 1: ωi
γ(B0  Gr  ri)

(4)

where ω i is the frequency of the proton at position ri and G is a vector representing the total gradient amplitude and direction. The dimensions of G are expressed in either milliTesla per meter (mT/m) or Gauss per centimeter (G/cm), where 1 G/cm
10 mT/m. Equation 4 states that, in the presence of a gradient field, each proton will resonate at a unique frequency that depends on its exact position within the gradient field. The MR image is simply a map of the frequencies and phases of the protons generated by unique magnetic fields at each point throughout the image. The displayed image consists of digital picture elements (pixels) that represent volume elements (voxels) of tissue. The pixel intensity is proportional to the number of protons contained within the voxel weighted by the T1 and T2 relaxation times for the tissues within the voxel.

Slice Selection The initial step in MRI is the localization of the rf excitation to a region of space, which is accomplished through the use of frequency-selective excitation in conjunction with a gradient known as the slice selection gradient, GSS. The gradient direction (x, y, or z) determines the slice orientation, while the gradient amplitude together with certain rf pulse characteristics determines both the slice thickness and slice position. A frequency-selective rf pulse has two parts associated with it: a central frequency and a narrow range or bandwidth of frequencies (typically 1–2 kHz). When such a pulse is broadcast in the presence of GSS, a narrow region of tissue achieves the resonance condition (Eq. 4) and absorbs the rf energy. The duration of the rf pulse and its amplitude determine the amount of resulting proton rotation (e.g., 90 degrees, 180 degrees). The central frequency of the pulse determines the particular location excited by the pulse when GSS is present. Different slice positions are excited by changing the central frequency (Fig. 2-17). The slice thickness is determined by the gradient amplitude GSS and the bandwidth of frequencies ∆ω incorporated into the rf pulse. Typically, ∆ω is fixed so that the slice thickness is changed by modifying the amplitude of GSS with thinner slices requiring larger values of GSS (Fig. 2-18). Once GSS determined, the center frequency is calculated using Eq. 4 to bring the desired location into resonance. Multislice imaging, the most commonly used approach for MRI, uses the same GSS but a unique rf pulse during excitation for each slice. Each rf pulse has the same bandwidth but a different central frequency, thereby exciting a different region of tissue. The slice orientation is determined by the particular physical gradient or gradients defined as the logical slice selection gradient. The slice orientation is defined so that the gradient direction (direction of magnetic field variation) is

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 41

Magnetic Resonance Imaging Principles and Applications

GSS ω3 ω2 ω1

∆z

∆z

∆z

z3

z2

z1

∆ω ∆ω ∆ω

Figure 2-17 Slice selection process. In the presence of a gradient (GSS), the total magnetic field that a proton experiences and its resulting resonant frequency depend on its position, according to Eq. 4. Tissue located at position zi will absorb rf energy broadcast with a center frequency ωi. Each position will have a unique resonant frequency. The slice thickness ∆z is determined by the amplitude of GSS and by the bandwidth of transmitted frequencies ∆ω.

perpendicular or normal to the surface of the slice, so that every proton within the slice experiences the same total magnetic field regardless of its position within the slice. Orthogonal slices are those in which only the x, y, or z gradient is used as the slice selection gradient. Oblique slices, those not in one of the principal directions, are obtained by applying more than one physical gradient when the rf pulse is broadcast. The total gradient amplitude determines the slice thickness. When images are viewed on the monitor or film, the slice selection direction is always perpendicular to the surface, that is, hidden from the viewer (Fig. 2-19). In MRI, the rf energy is applied as pulsed excitation, with several features characterizing the pulse. When a pulse is applied, there will be a center frequency, duration, shape, phase, and amplitude defined. The center frequency

41

of the pulse is normally chosen as the resonant frequency for the particular collection of protons under observation. The duration and shape of the pulse determine the bandwidth, or range, of frequencies on either side of the center frequency that is excited by the pulse. The phase of the pulse defines the effective orientation of the rf energy and determines the axis of rotation for the net magnetization under the influence of the pulse. The pulse amplitude determines the amount of rotation that the protons undergo (flip angle). In addition, the pulse amplitude is related to the amount of energy that the protons absorb and therefore must dissipate through T1 relaxation. In general, rf pulses are subject to two competing criteria: 1. Short-duration pulses require high peak pulse amplitudes to achieve the same pulse area (flip angle). Depending on the particular rf amplifier and transmitter coil, the maximum power that can be broadcast is limited. 2. For most applications, uniform amplitude and phase excitation throughout the tissue volume are desired; that is, all the protons within the slice should be rotated the same amount and in the same direction, producing a uniform excitation profile. This is only possible with low flip angles (less than 30 degrees). High-amplitude pulses, such as 90-degree or 180-degree pulses, have excitation profiles that are significantly nonrectangular. Manufacturers strive to provide uniform excitation profiles, subject to criterion 1. However, due to the finite nature of the rf pulse duration and amplifier power, the profile will not be rectangular, but will have sloped sides. This means that, for closely spaced slices, there will be areas that receive overlapping excitation, a problem known as crosstalk, between slices.

Readout or Frequency Encoding 4 mT/m

2 mT/m

∆ω

5 mm 10 mm

Figure 2-18 For a given range (bandwidth) of frequencies included in the rf pulse, the desired slice thickness is determined by the slice selection gradient amplitude. The user interface typically allows variation of slice thickness, which is achieved by increasing or decreasing the slice selection gradient amplitude, as appropriate.

The signal detection portion of the MRI measurement is known as the readout or frequency encoding. The readout process differentiates MRI from MR spectroscopy, the other type of MR process. In an imaging pulse sequence, the MR signal is always detected in the presence of a gradient known as the readout gradient GRO, which produces one of the two visual dimensions of the image on the film. A typical pulse sequence uses some form of excitation, such as a 90-degree slice selective pulse, to excite a particular region of tissue. Following excitation, the net magnetization within the slice is oriented transverse to B0 and will precess with frequency ω 0. T2* processes induce dephasing of this transverse magnetization (see the section titled Relaxation). This dephasing can be partially reversed to form an echo by the application of a 180-degree rf pulse, a gradient pulse, or both. As the echo is forming, the readout gradient is applied perpendicular to the slice direction. Under the influence of this new gradient field, the protons begin to

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 42

42

Chapter 2

Figure 2-19 Images in standard slice directions: sagittal, coronal, and transverse or axial. For transverse images, two view directions are possible: cranial and caudal. Annotations are based on patient axes.

precess at different frequencies depending on their position within it, in accordance with Eq. 4. Each of these frequencies is superimposed into the echo. At the desired time, the echo signal is measured by the receiver coil and digitized for later Fourier transformation. The magnitude of GRO (GRO) and the frequency that is detected enable the corresponding position of the proton to be determined (Fig. 2-20). Two user-definable parameters determine the magnitude of GRO: the field of view in the readout direction, FOVRO, and the Nyquist frequency ωNQ for the image, often referred to as the receiver bandwidth (Eq. 3). GRO is chosen so that protons located at the edge of the FOVRO precess at the Nyquist frequency for the image (Fig. 2-21). Smaller values of FOVRO are achieved by increasing GRO, keeping the Nyquist frequency, and thus the total frequency bandwidth, constant (Fig. 2-22). The spatial resolution, expressed as the voxel size with units of mm/pixel, is directly proportional to FOVRO and inversely proportional to the number of readout sample points in the acquisition matrix, NRO.

GR0

ω3

ω2 ω1

x3

x2

x1

Figure 2-20 Readout process. Following excitation, each proton within the excited volume (slice) precesses at the same frequency. During detection of the echo, a gradient (GRO) is applied, causing variation in the frequencies for the protons generating the echo signal. The frequency of precession ωi for each proton depends on its position xi according to Eq. 4. Frequencies measured from the echo are mapped to the corresponding position.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 43

Magnetic Resonance Imaging Principles and Applications

43

FIELD OF VIEW: READOUT

PHASEENCODING DIRECTION

FIELD OF VIEW: PHASEENCODING

Figure 2-21 In any image, one of

ω TR + ω NQ

ω TR

ω TR - ω NQ

READOUT DIRECTION

Phase Encoding The third direction in an MR image is the phase-encoding direction. It is visualized along with the readout direction in an image (see Fig. 2-21). The phase-encoding gradient, GPE, is perpendicular to both GSS and GRO and is the only gradient that changes amplitude during the data acquisition loop of a standard two-dimensional (2D) imaging sequence. Any signal amplitude variation detected from one

3mT/m

1.5mT/m

∆ω

250 mm 500 mm

Figure 2-22 For a given range (bandwidth) of frequencies that are measured in the signal, the desired field of view (FOV ) is determined by the readout gradient amplitude. The user interface typically allows variation of the FOV, which is achieved by increasing or decreasing the readout gradient amplitude, as appropriate.

the visualized directions is the readout direction, and the other is the phase-encoding direction. A proton located at the edge of the field of view in the readout direction (FOVRO) precesses at the Nyquist frequency (ωNQ) above or below the transmitter frequency ω TR. Changing the FOV of the image changes the spatial resolution (mm per pixel) but not the frequency resolution (Hz per pixel).

acquisition to the next is assumed to be caused by the influence of GPE during the measurement. The principle of phase encoding is based on the fact that the proton precession is periodic in nature. Prior to the application of GPE, a proton within a slice precesses at the base frequency ω 0. In the presence of GPE, its precessional frequency increases or decreases according to Eq. 4. Once GPE is turned off, the proton precession returns to its original frequency, but is ahead or behind in phase relative to its previous state. The amount of induced phase shift depends on the magnitude and duration of GPE that the proton experienced and the proton location. Protons located at different positions in the phase-encoding direction experience different amounts of phase shift for the same GPE pulse (Fig. 2-23). A proton located at the edge of the chosen FOV experiences the maximum amount of phase shift from each phase-encoding step. The MR image information is obtained by repeating the slice excitation and signal detection multiple times, each with a different amplitude of GPE. The second Fourier transformation in the image converts the signal amplitude at each readout frequency from a function of GPE to a function of phase. The spatial resolution in the phase-encoding direction depends on two user-selectable parameters, the field of view in the phase-encoding direction, FOVPE, and the number of phase-encoding steps in the acquisition matrix, NPE. The FOVPE is determined by the change in GPE from one step to the next. For a proton located at the chosen FOVPE, each phase-encoding step induces one half cycle (180 degrees) of phase change relative to the previous phase-encoding step, assuming a constant pulse duration (Fig. 2-24). NPE determines the total number of cycles of

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 44

44

Chapter 2 y3

y2

y1

GPE

φ1 φ2 φ3

During GPE

Before GPE

After GPE

φ1>0

y1

φ2=0

y2

y3

φ3BWREC

(8)

At 1.5 T, the frequency difference is 225 Hz so that, for a scan with a total receiver bandwidth of 20 kHz and NRO
256, the CSA will be 2.8 pixels. The severity of the artifact in the final image will depend on the FOVRO. If an FOVRO of 350 mm is used, this translates into a fat/water misregistration of 3.6 mm, whereas for an FOV of 175 mm, the artifact is only 1.8 mm. For a scan with a receiver bandwidth of 33 kHz, the CSA will be 1.7 pixels with a misregistration of 2.2 mm. CSAs are most prominent at magnetic field strengths

65

greater than 1.5 T, using low or narrow receiver bandwidths and large FOVs, and at fat–water tissue interfaces. In theory, CSAs are possible in all three directions (slice selection, phase encoding, and readout), since magnetic field variations (gradient pulses) are used for localization in all cases. In the slice selection direction, the rf pulse bandwidth and gradient amplitude are chosen to keep this at a minimum. In addition, because the slice selection direction is not directly visualized, any misregistration in this direction is difficult to discern. For routine imaging techniques such as spin echo, ETSE, or gradient echo, the phase-encoding process is reinitiated following each excitation pulse. Since ∆GPE is constant, fat and water protons located at the same position undergo equal amounts of phase change. They are mapped to the same location in the image and no artifact results. For these techniques, CSAs may be observed in the readout direction, based on the specific criteria described previously.

Phase Cancellation Artifact The second artifact induced by the chemical shift difference is known as the phase cancellation artifact, observed in outof-phase gradient echo images. As shown in Figure 2-11, the fat protons cycle in phase relative to the water proton precession. For normal spin echo and ETSE sequences, this phase cycling is exactly reversed by the 180-degree rf pulse(s) so that the fat and water protons always contribute to the signal at TE with the same polarity; that is, the fat and water protons are described as “in phase” when the signal is detected regardless of the choice of TE. In gradient echo sequences, the phase cycling is not reversed, and it causes fat and water protons to contribute differently to the detected signal, depending on the particular TE. Voxels containing both fat and water have additional signal intensity variations besides phase cycling. For certain choices of TE, known as “out-of-phase” TE, very little signal will be detected if the voxel has equal water and fat content, such as those voxels located at interfaces between fat- and water-based tissues. This signal cancellation appears as a dark ring surrounding the tissue (Fig. 2-58). The TE values that generate this phase cancellation depend on the magnetic field strength, since the time for the phase cycling depends on the resonant frequency difference in Hz between fat and water. The out-of-phase TE times occur midway between the in-phase times (Table 2-2). Out-of-phase images are often used to assess the amount of fat contribution to a voxel. The phase cancellation artifact observed using out-of-phase TE makes it difficult to assess the interface of tissues with different fat and water contents. The phase cycling also affects the signal content of all tissues throughout the image. Unless there is a specific reason for using an out-of-phase TE, such as for the assessment of fatty infiltration in the liver or adrenal masses or to use a very short TR, a TE corresponding to an in-phase relationship is preferable.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 66

66

Chapter 2

A

B Figure 2-58 Phase cancellation artifact. Other measurement parameters are: pulse sequence, two-dimensional spoiled gradient echo; TR, 164 ms; excitation angle, 70 degrees; acquisition matrix: number of phase-encoding steps, 134; number of readout sample points, 256; field of view, 263 mm in the phase-encoding direction  350 mm in the readout direction; NAV, 1; slice thickness, 9 mm. A: In-phase image (echo time [TE] 4.5 ms). Fat and water protons have the same phase and contribute in the same fashion to the image contrast. B: Out-of-phase image (TE 2.2 ms). Fat and water protons have opposite phases and contribute in opposite fashions to the image contrast. For voxels with equal amounts of fat and water, such as at the interface between liver or kidney and retroperitoneal fat, cancellation of signal occurs, producing a dark band (arrows).

Truncation Artifacts Truncation artifacts are produced by insufficient digital sampling of the echo. The most common instance of this condition occurs when data collection is terminated while significant signal is still being emitted by the protons, which can happen in T1-weighted imaging where there is a high signal from fat at the edge of the region still present at the end of data collection. The Fourier transformation of this truncated data set produces a “ringing” type of signal oscillation that emanates from the edge of the anatomy (Fig. 2-59). Truncation artifacts can also occur if the echo is sampled in an asymmetrical fashion; that is, the echo is not sampled equally on both sides of the echo maximum (Fig. 2-60). This type of data sampling is often used in gradient echo sequences with very short TEs ( 3 ms) to minimize the receiver bandwidth. Reduction of truncation artifacts involves reducing the signal amplitude so that it is minimal at the beginning or end of the data collection period. This reduction can be achieved in two ways. One is to reduce the fat signal

through fat suppression (since fat typically has the largest signal in T1-weighted images), either using a fat saturation pulse or using an IR technique. This approach changes the intrinsic image contrast and may be unacceptable for the particular clinical application. The other method is to apply an apodization filter to the raw data prior to Fourier transformation. This numerical process forces the signal amplitude to zero at the end of the data collection period.

TABLE 2-2 PHASE CYCLING TE TIMES Field strength, T

In-phase TE, ms

Out-of-phase TE, ms

0.5 1.0 1.5 3.0

13.3, 26.67 6.67, 13.3, 20 4.5, 9.0, 13.5, 18.0 2.25, 4.5, 6.75, 9.0

6.67, 20 3.3, 10, 16.67 2.25, 6.75, 11.25, 15.75 1.12, 3.38, 5.63, 7.88

Figure 2-59 Truncation artifact. Image acquisition with asymmetric sampling produces a banding artifact (arrows), originating from the high signal of subcutaneous fat. Measurement parameters: pulse sequence, spoiled gradient echo; TR, 170 ms; TE, 4 ms; excitation angle, 80 degrees; acquisition matrix: number of phaseencoding steps, 144; number of readout sample points, 256 with twofold oversampling; field of view, 262 mm in the phase-encoding direction  350 mm in the readout direction; NAV, 1; slice thickness, 8 mm.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 67

Magnetic Resonance Imaging Principles and Applications

b. B

a. A

TE Time

TE Time

Figure 2-60 Symmetric versus asymmetric echo sampling. A: In symmetric echo sampling, the echo is at a maximum (echo time, TE) midway through the sampling period so that both sides of the echo signal are measured equally. Filtering of the data can be performed in a symmetrical fashion. B: In asymmetric echo sampling, the echo maximum is in the early portion of the sampling period. Significant signal is present when the sampling begins and filtering of the signal is difficult.

The use of apodization filters improves S/N by removing high-frequency noise from the signal. However, excessive filtering eliminates high frequencies responsible for fine spatial resolution or edge definition so that image blurring is possible. Asymmetric echo sampling may require more extensive filtering to reduce truncation artifacts. Acquiring the echo in a symmetric fashion reduces the amount of filtering necessary.

Coherence Artifacts Coherence artifacts are a class of artifacts that can have a variable appearance, based on the particular measurement technique and how it is implemented on the scanner. They are produced by the rf pulses generating unwanted transverse magnetization that contributes to the detected signal. One type of coherence artifact may be produced as a result of multiple rf excitation pulses on the protons. The spin echoes used to produce T1- and T2-weighted images are but two of several echoes generated by rf excitation pulses within an imaging sequence. For example, a series of three rf pulses, such as a presaturation pulse and a 90-degree–180-degree pulse pair or a 90-degree–180-degree–180-degree pulse trio may generate four or five echoes. The timing and number of echoes depend on the exact spacing of the pulses. The amplitude of each echo depends on the excitation angle of the individual pulses and the particular tissues being measured. These “secondary” echoes contain all the frequency information of the so-called primary echoes normally used, but have different T1 and T2 weighting to the signals. Should these other echoes occur while the ADC is sampling the primary echoes, they will contribute to the final image. They may produce either line artifacts, if not phase encoded, or an additional image, if phase encoded.

67

Pulse sequences are usually designed with great care to minimize the contamination of the measured signals by the undesired echoes. When these undesired echoes form during the ADC sampling period due to the particular rf pulse timing, some form of coherence spoiling is used. Two approaches are used to dephase the transverse magnetization so that there is minimal coherence present from these secondary echoes. The most common method is to include additional gradient pulses applied at appropriate times during the pulse sequence, an approach known as gradient spoiling. These gradient pulses are usually of high amplitude and/or long duration. The time when these pulses are applied depends on which coherence is to be spoiled and what TE is chosen. Spoiler gradients applied following the 180-degree refocusing pulse reduce the FID artifact resulting from nonuniform rf excitation. Spoiler gradients following the data collection, such as those used in spin echo sequences (see Fig. 2-31), minimize coherence contamination of the subsequent echoes. For many sequences, the duration of the spoiler gradient pulse may be 30% to 50% of the slice loop. Gradient spoiling may also use a series of amplitudes rather than a constant amplitude pulse in order to improve spoiling if the rf interpulse time is short. The other approach for spoiling of unwanted transverse coherence is known as rf spoiling. Normal data acquisition techniques apply rf pulses of constant phase or with a 180degree phase alternation of the excitation pulse (i.e., 90 degrees, 90 degrees, 90 degrees,…). The receiver is also phase alternated to match the transmitter phase so that signal averaging can be performed. Rf spoiling applies a phase variation other than 90 degrees or 180 degrees to the excitation pulse and receiver. The desired echo signals add in a coherent fashion, while the other echoes add in an incoherent manner. If a sufficiently large number of phase variations is used, the net result is that the undesired signals average to zero. Rf spoiling does not require additional time during the sequence kernel, but does require sophisticated phase modulation of the transmitter.

Magnetic Susceptibility Difference Artifact A third artifact whose appearance is sensitive to the measurement sequence is caused by differences in magnetic susceptibility χ between adjacent regions of tissue. As mentioned in the first section, the magnetic susceptibility is a measure of the spin polarization induced by the external magnetic field. Tissues such as cortical bone or air-filled organs such as lungs or bowel contain little polarizable material and have very small values for χ. Soft tissue has a greater degree of polarization and larger χ. At the interface between soft tissues and the area of different susceptibility, a significant change in the local magnetic field is present over a short distance, causing an enhanced dephasing of the protons located there. As TE increases, there is more time for proton dephasing to occur, which results in signal loss. This dephasing is reversed by the 180-degree refocusing pulse in spin echo imaging since it is constant with

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 68

68

Chapter 2

Figure 2-61 Magnetic susceptibility difference artifact. As paramagnetic contrast agents accumulate in the kidneys, the local magnetic field is distorted, causing enhanced dephasing to the protons in the vicinity and producing signal voids (arrows). Measurement parameters are: pulse sequence, two-dimensional spoiled gradient echo; TR, 140 ms; TE, 4 ms; acquisition matrix: number of phaseencoding steps, 128; number of readout sample points, 256; field of view, 263 mm in the phase-encoding direction  350 mm in the readout direction; NAV, 1; slice thickness, 8 mm.

time, but contributes to the contrast in gradient echo imaging. Magnetic susceptibility dephasing is also observed following the administration of a paramagnetic contrast agent as the agent accumulates in the kidney and bladder. A significant signal loss occurs as the agent concentration in these organs increases (Fig. 2-61).

External Artifacts External artifacts are generated from sources other than patient tissue. Their appearance in the final images depends on the nature of the source and the measurement conditions. The sources may be classified into two general categories: those that originate from a malfunctioning or miscalibration of the MRI hardware and those that do not. Excluded from this group are hardware problems that cause complete failure of data collection or image reconstruction. Manufacturers exert great effort to ensure that the acquired images faithfully represent the MR signals from the defined area of interest. The artifacts described here are not specific to any particular manufacturer, but can occur on any MR system.

Magnetic Field Distortions One of the most common system-related problems is produced by constant distortions of the main magnetic field. During system installation, manufacturers perform a field optimization procedure known as shimming to eliminate coarse distortions of the base magnetic field caused by metal in the immediate vicinity of the magnet. However,

patients also produce field distortions due to their nonuniform shape and tissue content. These distortions can cause contrast variations within the image, particularly when fat saturation is used. Metal implants will distort the local magnetic field homogeneity surrounding them, producing significant artifacts. These artifacts frequently appear as an expansive rounded signal void with peripheral areas of high signal intensity distorting the surrounding regions, termed a “blooming” artifact. The size and shape of the artifact depends on the size, shape, orientation, and nature of the metal and the pulse sequence used for the scan (Fig. 2-62). Titanium or tantalum produces very localized distortions, whereas stainless steel can produces severe distortions that may compromise the image. The amount of image distortion observed from constant field inhomogeneities depends on the specific measurement technique. If the field homogeneity disruption is severe, the signal loss will be enough to preclude any image whatsoever. Gradient echo sequences are generally most sensitive to field distortions. This is due to the echo signal amplitude being a function of T2*, in which proton dephasing from magnetic field inhomogeneities is a significant factor affecting the image contrast. As in the case of magnetic susceptibility differences discussed earlier, very short TE values allow little time for such dephasing and result in smaller signal voids. Longer TE values allow more time for dephasing and can produce significant artifacts. Spin echo sequences, which include a 180-degree refocusing rf pulse, can also be used to minimize the intensity variations. ETSE techniques with very long echo trains are least sensitive to these distortions and should be used when metal is known to be in the imaging field. In addition, common metal objects within the bore, such as paper clips or staples, can cause severe image distortion. Another instance of magnetic field distortion occurs when fat saturation is used. As mentioned previously, fat saturation pulses are rf pulses centered at the fat resonant frequency that saturate the fat protons, leaving only the water protons to contribute to the image. If the field homogeneity is not uniform, the fat suppression pulse may not uniformly suppress the fat and may even suppress the water within the tissue. This condition results in regions of nonuniform fat suppression within the image (see Fig. 2-50B) and is most commonly observed at the edges of the optimized portion of the field, as can occur in images with large FOV or with extreme superior or inferior positions. It is advisable to try to center the anatomy within the magnet as much as possible and to perform a field homogeneity correction with the patient inside the scanner prior to fat saturation.

Measurement Hardware Hardware-induced artifacts are those produced by the malfunctioning of one or more of the scanner components during the data collection. Most MR techniques perform multiple measurements on the volume of tissue, varying only a

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 69

Magnetic Resonance Imaging Principles and Applications

69

A

B Figure 2-62 Magnetic susceptibility difference artifact. Surgical clips produce a void of signal caused due to the significant magnetic field distortion. A: Spoiled gradient echo sequence shows significant signal distortion (arrow). B: Single-shot echo train spin echo at the same level shows less signal loss (arrow).

single gradient amplitude (GPE) from one measurement to the next. The assumption is that any amplitude variation in the detected signal is caused by GPE, enabling localization in that direction. One of the primary requirements of the measurement hardware for this approach to succeed is that the gradient and rf transmitter systems act in a reliable and reproducible fashion. A lack of stability in the performance of either system causes amplitude or phase modulations or distortions in the detected signal in addition to those intrinsic to the measurement. These distortions result in smearing or ghosting artifacts in the phase-encoding direction throughout the entire image field. The magnitude and nature of the instability determines the amount of smearing. In many instances, the instability artifacts are indistinguishable from motion artifacts. Manufacturers perform tests during scanner calibration to assess the stability of the various components to ensure that their performance is reproducible and stable. Proper calibration of the measurement hardware is another important contribution to high-quality MR imaging. The gradient, rf transmitter, and receiver systems are calibrated to ensure their proper performance. Improper calibration produces variable distortions, depending on which component is considered. Nonlinear gradient pulses or incorrect amplitude calibration cause incorrect spatial localization and/or image distortion. Improper rf calibration causes incorrect or nonuniform excitation power, which may or may not be noticeable in the resulting images. The rf power deposition as measured by SAR will also be

inaccurate. Receiver miscalibration causes incorrect amplification of the echo signal, which may result in insufficient gain, so that the signal does not exceed the background noise, or excessive gain, which may cause the echo signal to exceed the digitization limits of the scanner (Fig. 2-63).

Noise A final artifact often present in MR images is noise. Noise can have a variety of appearances, depending on the origin and nature of the source. It may appear as a film superimposed over the normal anatomy, with or without discernible

Figure 2-63 Receiver miscalibration. The measured signal is amplified in excess of the maximum input voltage for the analogto-digital converter. Note the enhanced background signal (arrows).

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 70

70

Chapter 2

MOTION ARTIFACT REDUCTION TECHNIQUES

Figure 2-64 Spikes. Transient electrical discharges (spikes) during the data collection period produce a banding pattern that is superimposed across the entire imaging field. The direction and spacing of the bands depend on the timing of the discharge relative to the collection of the central phase-encoding steps.

patterns, or it may have a discrete pattern or patterns. The two most common examples of coherent noise are spikes and those arising from external sources. Spikes are noise bursts of short duration that occur randomly during the data collection. They are normally caused by static electricity discharges or arcing of electrical components, but may be generated by many different sources. Their appearance in an image depends on the severity, number, and location of the spike in relation to the signal maximum, but they tend to appear as waves superimposed on the normal image data (Fig. 2-64). They may or may not occur in all images of the scan. Spikes are particularly problematic to isolate because they are often irreproducible, particularly if the source is static discharge. External interference artifacts occur when there is a source of time-varying signal detected by the receiver. They appear as lines of constant frequency within the image. Their positions depend on the receiver bandwidth of the sequence and the frequency difference from the transmitter. The most common example of this is from the alternating nature (AC) of standard electrical current (60 Hz in the United States, 50 Hz in Europe and Asia). Electrical connections for any equipment, such as external patient monitoring devices, used in the scanner room should be filtered before being allowed to penetrate the Faraday shield, or the use of direct current (DC) should be considered. Manufacturers should be consulted before incorporating any electrical equipment into the scan room to ensure its compatibility with the MR scanner.

As discussed in the previous section, motion in an MR image can produce severe image artifacts in the phaseencoding direction. The severity of the artifact depends on the nature of the motion, the time during data collection when the motion occurs, and the particular pulse sequence and measurement parameters. The most critical portion of the data collection period for artifact generation is during the collection of echoes following low-amplitude phaseencoding steps (GPE
0 mT/m). Motion during the highamplitude phase-encoding steps (GPE or

0 mT/m) can cause blurring but not severe signal misregistration in the image. Four methods are commonly used to reduce the severity of the motion artifact on the final image. If signal from the moving tissue is not of interest, use of a spatial presaturation pulse can significantly reduce artifactual signal. This is useful to suppress blood flow from the aorta or inferior vena cava in abdominal examinations. The other techniques are useful when signal from the moving tissue is desired. Two of these, acquisition parameter modification and physiological triggering, affect the mechanics of the data collection process, while the third method, flow compensation, alters the intrinsic signal from the moving tissue. Although none of these approaches completely removes motion artifacts from the image, use of one or more of these techniques substantially reduces the impact of motion on the final images.

Acquisition Parameter Modification Proper choice of the acquisition parameters can alter the appearance of motion artifacts. One example of this is to define the phase-encoding direction so that the motion artifact does not obscure the area of interest. This may be referred to as motion artifact rotation, or swapping the frequency and phase-encoding directions. This approach does not eliminate or minimize the artifact, but only changes its position within the image. For example, motion artifacts due to blood flow from the aorta or inferior vena cava may obscure lesions in the left lobe of the liver if the phase-encoding direction for a transverse slice is in the anterior–posterior direction. If the phase-encoding direction is in the left–right direction, the artifact will be present in the right lobe. Alternatively, removal of respiratory artifacts in abdominal or cardiac imaging can be accomplished by measuring all the scan data within one breath-hold. For example, a TR
140 ms, NPE
128, and NSA
1 can produce a complete scan in 18 seconds. The spatial resolution may be compromised in the phase-encoding direction, depending on the FOV, but respiratory motion and its artifacts will not be present if the patient suspends respiration. Extreme examples of this approach are single-shot ETSE and EPI

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 71

Magnetic Resonance Imaging Principles and Applications

Figure 2-65 Single-shot echo train spin echo image of the abdomen. Note lack of motion artifact from heart and bowel.

techniques, where images can be acquired in less than 1 second, producing heart and bowel images that are virtually motion-artifact free (Fig. 2-65).

Triggering/Gating Another method for visualizing moving tissue allows the tissue to move but synchronizes the data collection with a periodic signal produced by the patient such as a pulse or heartbeat. The data collection–timing signal relationship can be exploited in two ways. Prospective methods initiate the data collection following detection of the timing signal. The signal detection for a particular slice always occurs at the same time following the timing signal. Since the moving tissue is in the same relative position at this time, there is minimal misregistration of signal and a significant reduction of the resulting motion artifact. Retrospective methods measure the timing signal together with the echo signal, but the timing signal is not analyzed nor the measured data adjusted until the scan is completed. The data collection process can also be triggered by or gated to the timing signal. Triggering examinations are commonly prospective in nature and begin the data collection following detection of a signal. The noise produced by the scanner during sequence execution is also synchronized to the signal source. The TR controlling the contrast is based on the repetition time for the trigger signal TREP rather than the TR entered in the user interface. Several external methods for detection of the timing signal can be used. For cardiac imaging, data collection is synchronized to the ECG signal measured from the patient using lead wires. Pulse triggering

71

uses a pulse sensor to detect the pulse, usually connected to an extremity. Respiratory triggering uses a pressure transducer attached to the patient to measure either chest or abdominal motion. An internal method for a timing reference uses a so-called “navigator echo” that, when properly positioned, can measure an MR signal from the diaphragm or other tissue and indicate that motion has occurred. There are several potential problems with triggered studies. One is that the scan time is longer for a triggered study than for the corresponding untriggered study. Time must be allowed from the end of data collection (traditionally TR) to the next trigger signal to ensure that the trigger signal is properly detected by the measurement hardware. This extra time is usually 150 to 200 ms to allow for variation of the rate of the trigger signal for the patient. The total scan time will be extended by approximately 1 to 2 minutes. Another problem occurs if the rate of the trigger signal is irregular. The TREP for a particular phase-encoding step depends on the R–R time interval immediately prior to the step under consideration. Variation in the heart rate causes variation in the amount of T1 relaxation from measurement to measurement for each phase-encoding step. This variation produces amplitude changes in the detected signal, producing misregistration artifacts in the final images even if the triggering is perfect. Stability of the heart rate is most critical during collection of the echoes with low-amplitude GPE (GPE
0 mT/m). Cardiac examinations are one example of an examination that typically uses prospective triggering. The peak R wave is normally used as the timing reference point. Each phase-encoding step is acquired at the same point in time following the R wave so that the heart is in the same relative position (Fig. 2-66). Since the kernel time per slice is shorter than the duration of the R–R interval (TREP), several signals can be acquired within one heartbeat. Proper detection of the ECG signal from the patient is critical. Improper electrode placement may detect significant signal from the blood during its flow through the aortic arch. In addition, the transmitted rf energy and the gradient pulses may also interact with the lead wires, inducing significant noise to the detected ECG signal. High-resistance lead wires may even burn the patient through this coupling. Two types of examinations are commonly used in cardiac studies. Spin echo imaging acquires one signal from a slice per heartbeat. Each image within the scan is acquired at a different position and a different time point in the cardiac cycle. These images are typically T1 weighted (depending on the R–R time interval) with minimal blood signal and are used for morphological studies of the heart. Another approach produces significant blood signal and is known as cine heart imaging. In this method, multiple images are acquired at the selected slice position during the R–R time interval. Rapid display of these images allows a dynamic visualization of the heart and cardiac hemodynamics during the different phases of the cardiac cycle

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 72

72

Chapter 2

R-R

TD

TD

TD

TD

Multislice SinglePhase 1 2 3 4 5 6

1 2 3 4 5 6

1 2 3 4 5 6

1 2 3 4 5 6 Slice#

1 2 1 2 1 2

1 2 1 2 1 2

1 2 1 2 1 2

1 2 1 2 1 2 Slice#

Multislice Multiphase (Cine)

(Fig. 2-67). Traditional cine heart examinations use gradient echo sequences with sequential filling of the raw data matrix, acquiring one echo per slice per heartbeat. However, more advanced cine examinations use a segmented filling, allowing multiple echoes per slice to be acquired per heartbeat. Unlike the ETSE sequence, in which multiple

Figure 2-66 Principle of electrocardiogram (ECG) triggering. Triggering the data collection to the ECG signal from the patient decreases the severity of the motion artifact. The R wave is used as an initiation signal. For multislice, singlephase imaging, one phase-encoding step for each slice is acquired per heartbeat. Information for a particular slice is acquired at the same time following the R wave. The T1 contrast is based on the R–R time interval rather than the user-defined repetition rate. Alternately, a multislice, multiphase acquisition may be performed in which slices are acquired at different times during the cardiac cycle at the same position. A trigger delay (TD) can be used to initiate data collection at any desired time during the cardiac cycle.

echoes are measured following a single excitation pulse, segmented gradient echo techniques repeat the gradient echo kernel multiple times, each repetition with a different GPE amplitude. This allows a complete raw data set to be acquired in fewer heartbeats, which reduces the scan sensitivity to a variable heart rate. In many cases, the scan times

A

B Figure 2-67 Cine heart images acquired at different phases of the cardiac cycle. Measurement parameters: pulse sequence, two-dimensional spoiled gradient echo; TR, 35.4 ms; TE, 1.5 ms; excitation angle, 65 degrees; acquisition matrix: number of phase-encoding steps, 108; number of readout sample points, 128; field of view, 169 mm in the phase-encoding direction  200 mm in the readout direction; NAV, 1; slice thickness, 6 mm. (A) 92 ms following R wave. (B ) 552 ms following R wave.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 73

Magnetic Resonance Imaging Principles and Applications

are short enough for patients to suspend respiration, which reduces artifacts from respiratory motion. Prospective cine examinations have a problem in that the initial image in the dataset, representing the first phase of the cardiac cycle for the scan, often is brighter than the subsequent images in the scan. This effect results from the longer amount of T1 relaxation that the net magnetization undergoes for this time point compared with the other images, due to the “dead” time waiting for the trigger pulse. Gating methods typically have the sequence continuously executing rf and gradient pulses but not measuring signal, and use the timing signal to control when echoes are actually measured. The continuous execution maintains a relatively constant net magnetization so that all images will have comparable contrast. Prospective gating has been used in cardiac imaging and in abdominal imaging to reduce artifacts from respiratory motion. Two approaches are used, both of which synchronize the data collection to the respiratory cycle of the patient. Simple respiratory gating acquires the data when there is minimal motion. It suffers from significantly longer scan times since the time during respiratory motion is not used for data collection. An alternative technique used for T1weighted imaging, respiratory compensation, rearranges the order of phase encoding so that adjacent GPE are acquired when the abdomen is in the same relative position. Typically, the low-amplitude GPE are acquired at or near end expiration so that the echoes contributing the most signal are measured when there is the least motion. Higher amplitude GPE are acquired during inspiration. Significant improvement of respiratory-induced ghosts can be achieved by either technique as long as there is a uniform respiration rate during the scan. Nonuniform respiration may produce artifacts as severe as those produced from a nongated scan. An alternative to prospective triggering for cine heart imaging is known as retrospective gating. In this approach, the ECG signal is measured but the data collection is not controlled by the timing signal. Instead, the data are measured in an untriggered fashion and the time following the R wave when each phase-encoding step was measured is stored with it. Following completion of the data collection, images are reconstructed corresponding to various time points within the cardiac cycle. The data for any phase-encoding step not directly measured are interpolated from the measured values.

Flow Compensation A final method for reducing motion artifacts adds additional gradient pulses to the pulse sequence to correct for phase shifts experienced by the moving protons. It is known as flow compensation, gradient motion rephasing (GMR), or motion artifact suppression technique (MAST). As described previously, gradient pulses of equal area but opposite polarity are used to generate a gradient echo.

73

Proper dephasing and rephasing of the protons and correct frequency and phase mapping occur as long as there is no motion during the gradient pulses. Movement during either gradient pulse results in incomplete phase cancellation or a net phase accumulation at the end of the second gradient pulse time. The amount of phase accumulation is related to the velocity of the motion. This phase accumulation produces signal intensity variations that are manifest as motion artifacts in the phase-encoding direction (Fig. 2-68), regardless of the direction of the motion or the gradient pulses. If the motion of the protons is relatively simple with respect to time, then the motion-induced phase changes can be predicted and may be corrected by applying additional gradient pulses. These pulses are applied in the direction for which compensation is desired. The number, duration, amplitude, and timing of the pulses can be defined so that protons moving with constant velocity (first-order motion), acceleration (second-order motion), and pulsatility (third-order motion) can be properly mapped within the image. Flow compensation requires that certain limitations be placed on the pulse sequence. Because additional gradient pulses are applied during the slice loop, the minimum TE for the sequence must be extended to allow time for their application. In pulse sequences where short TEs are desired, higher amplitude gradient pulses of shorter duration may be used. This will limit the minimum FOV available for the sequence. In practice, only modest increases in TE and the minimum FOV are normally required.

MAGNETIC RESONANCE ANGIOGRAPHY One of the strengths of MRI for imaging patient tissue is its ability to acquire information regarding its function as well as its structure. The most common example is the examination of flowing blood within the vascular network using MR angiography (MRA). While moving tissue can produce severe image artifacts, MRA uses flowing tissue, such as blood, to provide the primary source of signal intensity in the image. It provides visualization of the normal, laminar flow of blood within the vascular system and its disruptions due to pathologic conditions such as stenoses or occlusions. MRA can be of particular benefit in evaluating vessel patency. MRA techniques have the advantage over conventional x-ray–based angiographic techniques in that use of a contrast agent is not always required. Consequently, multiple scans may be performed if desired (e.g., visualizing arterial and then venous flow). The most common MRA approach is known as a “bright blood” method. Signal from the moving protons is accentuated relative to the stationary protons through the pulse sequence and measurement parameters. Whether the vessels are displayed as white pixels on a dark background or dark pixels on a bright background, bright blood MRA

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 74

74

Chapter 2

B

A Figure 2-68 Flow compensation. Use of flow compensation gradient pulses will map moving protons, such as those in cerebrospinal fluid (CSF), to their proper locations. Measurement parameters are: pulse sequence, spin echo; TR, 2,500 ms; TE, 90 ms; excitation angle, 90 degrees; acquisition matrix: number of phase-encoding steps, 192; number of readout sample points, 256 with twofold readout oversampling; field of view, 210 mm in the phase-encoding direction  280 mm in the readout direction; NAV, 1; slice thickness, 5 mm. A: No flow compensation. Misregistration artifact from CSF flow appears anterior to the spinal canal (arrow). B: First-order flow compensation in readout and slice selection directions. CSF is properly mapped into the spinal canal.

techniques assign high pixel amplitudes to the laminarflowing blood. Bright blood MRA techniques can be divided into two approaches: time-of-flight and phase-contrast methods. In their simplest implementation, both methods visualize arterial and venous flow simultaneously through the volume of interest. This can lead to ambiguous identification if the vessels are in close proximity to each other. One way to select either arterial or venous flow is to apply spatial presaturation pulses to saturate the undesired blood signal prior to its entry into the imaging volume (Fig. 2-69). Presaturation pulses may also be used in “black-blood” MRA techniques, in which the blood signal is saturated and the vessels appear dark relative to the surrounding tissue. These techniques are seldom used and will not be discussed further in this book. MRA techniques use gradient echo sequences as the measurement technique for data collection and may use 2D sequential slice (number of slices equal to the number of subloops) or 3D-volume acquisition modes. Gradient echo sequences allow the use of short TE times (usually less than 10 ms), which minimize T2* dephasing of the blood signal. The choice of 2D versus 3D is usually dictated based on the total volume of tissue to be examined and the total scan time. Three-dimensional techniques

provide the best in-plane resolution and minimize vessel misregistration artifacts. However, saturation of blood at the distal side of the volume limits the slab thickness to approximately 70 mm unless a T1 relaxation agent is used. 3D methods require significantly longer scan times. The 2D method is preferable for imaging vessels with slow flow velocity, such as veins, or to limit saturation effects from in-plane flow. Coverage of large areas of anatomy is A

imaging volume

B presaturation imaging region volume

Figure 2-69 Flow selection in magnetic resonance angiography. A: Blood flow from both directions into the imaging volume will be visualized in the image. B: Positioning of a presaturation pulse to saturate blood before it enters the imaging volume will allow only unsaturated flow to be observed.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 75

Magnetic Resonance Imaging Principles and Applications

possible in one scan since the volume of excitation is typically 3 to 4 mm per slice. There are two additional problems associated with MRA. One is that the saturation pulse becomes less effective as more time elapses between the presaturation and the excitation pulses. This problem occurs because of both increased T1 relaxation of the blood and time-of-flight effects (see next section). This loss of effectiveness can be minimized by proper positioning of the presaturation pulse near the volume of excitation. Care must be taken in positioning if there is pulsatile flow in the vessel of interest, as too close a position to the imaging volume could cause a reduction in signal if the flow is retrograde. A second problem is an exaggerated sensitivity to vessel stenosis. The stenotic region disrupts the laminar flow in the area of and distal to the stenosis, causing a loss of vessel signal to a greater extent than from the stenotic region alone. As the laminar flow is reestablished within the vessel, bright signal can again be measured.

Time-of-Flight Magnetic Resonance Angiography Time-of-flight techniques are the most time-efficient methods for obtaining MRA images. A single measurement is performed, with the stationary tissue signal suppressed relative to the flowing tissue signal. A moderate excitation pulse angle and a TR much shorter than the tissue T1 values are used to accomplish this. While the stationary tissue experiences every rf excitation pulse, the flowing tissue does not. A volume of blood will be at a different location at the time of each excitation pulse due to its motion during TR (Fig. 2-70). The signal from the blood volume is largest at the entry point of the slice because it has not TR

B

A

TR

TR

C

D

Figure 2-70 Time-of-flight effect. During data collection, the imaging volume experiences multiple radiofrequency (rf) pulses. Flowing blood (shaded box) experiences the first rf pulse (A). During the first repetition time (TR) period, the excited blood volume moves (B) and only a portion experiences the second rf pulse. During the second TR period, the initial volume of blood continues to move (C). By the end of a third TR period, the initial volume of blood is entirely outside the volume of excitation and does not contribute to the detected signal (D).

75

undergone any excitation pulses. As the blood volume travels through the slice, it becomes saturated as it undergoes more excitation pulses and loses signal. If the flow direction is perpendicular to the slice (through-plane flow), and the volume of excitation is small, then the volume of blood exits the slice before it is completely saturated and significant blood signal can be measured throughout the entire slice. The degree of blood saturation depends on the slice thickness, TR, excitation angle, and flow velocity. Time-of-flight techniques suffer from incomplete suppression of the stationary tissue, particularly fat, due to the faster T1 relaxation times of the stationary tissues relative to the blood. Three-dimensional time-of-flight techniques also have increased saturation of the blood due to the large number of excitation pulses of the imaging volume. This may be overcome by the administration of T1 contrast agents to shorten the T1 relaxation time of blood and obtain more blood signal (Fig. 2-71). Measurement times can be reduced to a few seconds through the use of interpolation in the slice selection, enabling serial scan volumes to be acquired coupled with suspension of respiration during the scan. Subtraction of pre- and postcontrast volumes eliminates residual signal from stationary tissue. Use of contrast agents for abdominal MRA has enabled rapid, reliable, and high-quality studies to be obtained with minimal contamination from patient motion. Studies of peripheral vessels by MRA can also be performed following contrast administration. Multilevel acquisitions together with rapid movement of the patient table allow for run-off studies of extremities within a single short imaging session.

Phase-Contrast Magnetic Resonance Angiography Phase-contrast MRA is a technique in which background tissue signal is subtracted from a flow-enhanced image to produce flow-only images, analogous to digital subtraction x-ray angiography. The acquisition sequence may produce a one-dimensional profile or standard 2D or 3D images. A minimum of two images is measured at each slice position. One image, known as the reference image, is acquired with flow compensation. Subsequent images are acquired following application of additional gradient pulses that induce a phase shift in blood moving with a particular direction and flow velocity. The moving protons are rephased at TE, and the resultant image is subtracted from the reference image to produce images of only the moving protons. The blood signal amplitude depends on its velocity, with maximum signal obtained from flowing blood at an operator-specified velocity known as the venc, or velocity encoding value. Additional scans may be performed to sensitize flow in other directions or at other velocities. The resultant difference images may be combined to produce images sensitive to flow in any direction at the chosen velocity.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 76

76

Chapter 2

A

B Figure 2-71 Magnetic resonance angiography of aorta and renal arteries following bolus administration of gadolinium-chelated contrast agent. Measurement parameters are: pulse sequence, three-dimensional refocused gradient echo, postexcitation; TR, 3 ms; TE, 1.1 ms; excitation angle, 20 degrees; acquisition matrix: number of phase-encoding steps, 128; number of readout sample points, 256; field of view, 350 mm in the phase-encoding direction  400 mm in the readout direction; NAV, 1; effective slice thickness, 2.0 mm. A: Coronal source image, one from data set. Image was obtained following subtraction of enhanced scan from unenhanced scan. B: Coronal postprocessed image of (A).

Phase-contrast MRA has several advantages over timeof-flight techniques. Being subtraction rather than saturation techniques, they have better background suppression than time-of-flight methods. Through their directional sensitization, phase-contrast methods enable flow in each primary direction to be separately assessed. In addition, quantitation of the flow velocity in each direction is possible. However, phase-contrast methods suffer from two problems. One is a significantly longer scan time. Four separate acquisitions are required to measure flow in all three directions. The scan time is four times longer than for a time-of-flight method with similar acquisition parameters. In addition, prior knowledge of the maximum velocity is necessary to ensure that the proper venc is used. The maximum signal is achieved for blood flowing at the

chosen venc. Flow velocities exceeding this velocity are misregistered in the image and appear as slower flow, a situation known as flow aliasing. This artifact is analogous to the aliasing problem observed in phase encoding (see section on phase encoding). Blood signal is reduced at these higher velocities until a complete loss of flow signal occurs at flow velocities that are even multiples of the venc. At the other extreme, a venc too large will minimize contrast between velocity changes within a vessel. Proper choice of the venc allows adequate visualization of all flow with good sensitivity to velocity variations within a vessel. This choice can be facilitated by the use of velocity measurements using Doppler ultrasound techniques or by scanning the volume using 2D techniques with multiple vencs to see which one provides the best results.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 77

Magnetic Resonance Imaging Principles and Applications

Maximum-Intensity Projection Regardless of the choice of acquisition technique, examination of the individual images from an MRA scan can provide details regarding flow variations within a vessel. However, a vessel is seldom located within a single slice but usually extends through several slices at an arbitrary angle, making the vessel tortuosity or spatial relationship with neighboring vessels difficult to assess, particularly if the vessels are oriented perpendicular to the imaging volume. Bright-blood MRA images may be analyzed using a postprocessing technique known as maximum-intensity projection (MIP) to better visualize the 3D vessel topography. The MIP process generates images from the entire set of MRA images. A view direction is chosen, and the entire set of images is projected along that direction onto a perpendicular plane using a “ray tracing” approach. The pixel of maximum intensity is chosen as the pixel for the MIP image, regardless of the input slice where the pixel is located (Fig. 2-72). Because the bright-blood MRA technique accentuates blood signal over stationary tissue signal, the MIP process preferentially selects blood vessels whenever possible, which enables the entire vessel to be examined no matter where it is located within the imaging volume. Multiple images may be obtained from the same data set through change of the view direction (rotation of the projection angle). Vessels that are superimposed in one projection may be clearly resolved in another one. It is also possible to perform the MIP process on a subset of the data, a so-called “targeted” approach. This is useful for tailored reconstruction of the renal arteries. Care must be taken in the definition of the targeted area that the vessel of interest does not leave the area. The MIP process causes the vessel to be cut off at the edge of the defined region, simulating a vessel obstruction (Fig. 2-73). Careful exami-

A

B

Figure 2-72 Maximum-intensity projection (MIP). The magnetic resonance images are acquired so that moving blood has pixels of maximal intensity. The MIP process maps the pixels of maximum intensity into a single projection or view regardless of which slice the pixel was located in. Changing the direction of projection provides a different perspective of the vessels.

77

nation of the source images and/or MIP images in multiple orientations is necessary to ensure the inclusion of the entire vessel within the reconstructed area.

ADVANCED IMAGING APPLICATIONS Initially, MR imaging was restricted to visualizing hydrogen nuclei using basic spin echo techniques. As hardware and software have evolved, newer and faster imaging techniques have been developed. The advent of subsecond imaging techniques has provided the ability to measure metabolic processes within tissues with significantly faster temporal resolution than was previously possible. Two areas where these advances have been applied to body imaging are studies of tissue perfusion and the use of ultrahigh magnetic fields.

Perfusion Although angiographic techniques visualize the vascular network within a patient, they do not have sufficient spatial resolution to visualize blood flow through a tissue in bulk. However, it is possible in many instances to observe changes in tissue signal due to the blood flow through it, a process known as perfusion. Proper tissue perfusion is critical to ensure an adequate supply of nutrients to the constituent cells as well as removal of metabolic byproducts. It also aids in maintenance of a stable tissue temperature. Abnormalities in perfusion can lead to an increased temperature sensitivity and a loss of tissue viability through hypoxia. Three approaches are used for MR perfusion studies in the abdomen, which are analogous to radioisotope tracer studies and use similar methods for analysis of the flow dynamics. One approach acquires a series of rapid ( 20 seconds per image) T1-weighted imaging studies following the bolus administration of a gadolinium-chelated contrast agent. These images are typically acquired using spoiled gradient echo, T1-weighted MP or echo planar techniques. An increase in tissue signal occurs as the contrast agent infuses the extravascular spaces of the tissue. Perfusion defects are visualized as a lack of signal increase for the affected region of tissue. A second approach is useful if the contrast agent remains in the blood vessels. In this case, the paramagnetic nature of the contrast agent increases the local tissue susceptibility, causing increased T2* dephasing of nearby tissues. Serial T2*-weighted gradient echoes are acquired, and the well-perfused tissue has a reduction of signal relative to the precontrast images or the poorly perfused tissues. The third approach does not use contrast media to highlight the flowing spins. Instead, two sets of images are acquired using an EPI pulse sequence. One set is acquired following a region-selective inversion pulse that inverts or “tags” the spins outside the slice of interest, while the

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 78

78

Chapter 2

B

A Figure 2-73 Maximum-intensity projections (MIPs) of the magnetic resonance angiography exam from Figure 2-71A, coronal (A) and sagittal (B) projections. Note absence of vessel in (A) indicated by arrow is due to its exclusion in the MIP, as indicated by arrow in (B).

other set of images serves as a reference. A variable delay time between the tagging pulse and the data collection allows control of the tag position based on the flow velocity. Two flow imaging methods have been developed using this tagging approach that differ in the nature of the reference image. One method is known as EPISTAR (1). The reference image is acquired following application of either an inversion pulse on the opposite side of the imaged slice or a pair of inversion pulses at the same position as the tagging inversion pulse. The other approach is known as FAIR (2). This method acquires the reference image following a non–slice-selective inversion pulse. In both methods, the resulting difference image has signal primarily from the moving spins. Two examples where perfusion studies have shown promise are in the examination of abnormalities of blood flow within tissue and for the detection of tumors. Blood flow anomalies in the myocardium following infarction have been studied for many years using radionuclide agents. First-pass MR perfusion studies have shown good correlation with these studies and have enabled visualization of different phases of perfusion (8). In these examina-

tions, the measurement parameters are chosen so that the blood signal is minimal prior to contrast agent administration. Reduced uptake of the contrast agent in poorly perfused tissue causes a delay in signal enhancement. Liver studies following administration of gadolinium-chelated contrast agents have also demonstrated differences in tissue perfusion. Images acquired immediately following contrast administration show capillary phase perfusion, whereas images acquired 45 seconds postadministration show substantial portal phase perfusion (Fig. 2-74). The normal spleen usually shows a serpigenous enhancement pattern immediately following contrast administration, with a more uniform signal intensity observed on images acquired after 45 seconds or later. The other studies where differential perfusion has been used are in the detection of certain tumors. Malignant breast tumors have demonstrated a significantly faster uptake of contrast media than benign tumors in some studies (3). Rapid scanning is necessary because both classes of tumors have similar signal amplitudes 3 minutes following contrast administration. Use of a 3D volume scan enables a bilateral breast study with good spatial and temporal resolution.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 79

Magnetic Resonance Imaging Principles and Applications

79

B

A Figure 2-74 T1-weighted spoiled gradient echo imaging of the liver following the administration of a gadolinium-chelated contrast agent. A: Image acquired immediately following administration. Contrast agent is in the hepatic arterial phase, as evidenced by the nonopacified hepatic vein (arrow) and serpigenous enhancement of the spleen. B: Image acquired 45 seconds following administration. Contrast agent is now in the capillary phase, as evidenced by its presence in the hepatic vein (arrow) and uniform signal from the spleen.

Ultrahigh Field Imaging* The development of MR scanners with B0 significantly greater than 1.5 T presents a number of challenges. While clinical images have been demonstrated at field strengths of 8.0 T, commercially available scanners are limited to 4.0 T or less. The challenges of so-called ultrahigh magnetic field scanners can be divided into two categories: those exclusive of patient tissue and those due to changes in the tissue response to the increased magnetic field. As might be expected, many of the areas of concern with conventional field strength magnets are also valid with ultrahigh field systems, and in some cases of greater concern. For example, current ultrahigh field magnets are superconducting solenoidal magnets and use liquid helium as the cryogen, as are most conventional magnets. The higher field strengths require greater amounts of magnet wire windings, increasing the overall size and weight of the magnet, and greater amounts of electrical current to generate the magnetic field. The larger B0 exerts larger forces of attraction than conventional magnets of the same physical size. This increases the audible noise of the scanner during measurements. The fringe magnetic field extends outside the magnet housing to a greater extent, increasing the influence of the environment on the magnet homogeneity. Shimming of the magnet is also more problematic, in that the absolute inhomogeneity (measured in Hz or µ T over a particular distance or volume) increases with increased B0, while the relative homogeneity (measured in ppm) might remain constant. Ferromagnetic objects must be kept at greater

*This section discusses concepts that are more fully explained in the section on instrumentation.

distances to prevent uncontrollable attraction into the bore. Safe distances for patients with metal implants or for electronic equipment extend farther from the magnet isocenter. Proper site planning and preparation is critical for safe and successful magnet installations. As with scanning at conventional B0, there are issues with rf power at ultrahigh field strengths. As stated in Eq. 1, the resonant frequency ω is proportional to B0. There are greater problems with rf penetration into patients at higher frequencies, a phenomenon known as the skin effect. This can cause inhomogeneous excitation of a slice, producing shading in images. In addition, more power from the transmitter is necessary to produce the rf pulses. As mentioned in the instrumentation section, the power from a pulse is used in the calculation of the SAR for a scan. The power produced by a pulse is proportional to the square of its frequency, so that increasing B0 by a factor of 2 increases the pulse power by a factor of 4. This means that whole-body scanning at ultrahigh B0 requires different examination protocols so as not to exceed current SAR guidelines. Additional tissue response issues can have a significant impact on clinical scanning. The T1 relaxation times increase significantly with increasing B0, whereas the T2 relaxation times are relatively constant. This means that significant T1 saturation can occur in scans at ultrahigh B0 using measurement parameters that produce minimal saturation with conventional scanners. As a result, longer TR times are necessary to achieve equivalent tissue contrast. This aspect, together with the increased SAR, has limited use of traditional spin echo for acquiring T1 images. Instead, use of gradient echo is being explored to provide acceptable image quality. There are also greater magnetic

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 80

80

Chapter 2

susceptibility differences at ultrahigh B0. This can produce more severe artifacts in areas where these differences are significant, but can be exploited when performing perfusion MR examinations.

INSTRUMENTATION A very important aspect of the production of MRI images is the instrumentation used in the measurement. Many MR systems are commercially available, each possessing different features and capabilities that are often difficult to evaluate and compare objectively. These features are often determined by the operating software provided by the manufacturer, but certain hardware components are common to all systems. The following sections describe the basic subsystems of an MRI scanner and technical aspects to consider when comparing systems. The major components are a computer/array processor, a magnet system, a gradient system, an rf system, and a data acquisition system (Fig. 2-75).

Computer Systems Currently, every MRI system has a minimum of two computers. The main, or host, computer controls the user interface software. This software enables the operator to control all functions of the scanner, either directly or indirectly. Scan parameters may be selected or modified, patient images may be displayed or recorded on film or other media, and postprocessing such as region-of-interest measurements or magnification can be performed. Several peripheral devices are typically attached to the main computer. One or more hard disks are used to store the patient images immediately following reconstruction. Such disks are used for short-term storage; current disk drives are capable of storing 100,000 to 150,000 images, depending on the image size. A device for long-term archival storage, either a laser optical disk or a CD-ROM, is usually included.

GRADIENT AMPLIFIER

GRADIENT COIL

HOST COMPUTER

RF AMPLIFIER

TRANSMITTER COIL

IMAGE PROCESSOR

RECEIVER

RECEIVER COIL

SHIM SYSTEM

MAGNET

Figure 2-75 Block diagram of magnetic resonance imaging system. RF, radiofrequency.

A hard-copy camera may be connected via a network or attached to the main computer if filming is controlled by the operating software. The second computer system is a dedicated image processor used for performing the Fourier transformations or other processing of the detected data. This processor is controlled by software provided by the manufacturer. The raw data are stored from the receiver into memory in the image processor itself or onto a separate hard disk. The image processors currently used with MRI scanners are capable of performing the Fourier transformation for a 256  256 matrix in less than 50 milliseconds. Additional image processors may be present to perform more computationally intensive postprocessing. One or more consoles may be attached to the main computer or may have direct access to the reconstructed images. The main console is the primary device for operator input. Each console has a keyboard and one or more monitors for displaying images and text information. A computer mouse is often used for more interactive control. Additional consoles may be directly attached to the same main computer, allowing convenient viewing or postprocessing of the images. Alternately, other viewing stations may be completely detached from the primary computer system and access the image data through a network connection. They may be used as a common viewing station for analysis of images from additional imaging modalities, such computed tomography (CT), ultrasound, or traditional radiography. In addition, there may be an archive server known as a picture archiving and communications system (PACS) unit. This system is used as a centralized digital repository for images. PACS units allow multiple users to access the images and patient data easily and can provide long-term archiving of the images in a central location. MRI systems are often incorporated into a computer network for the facility. This allows images to be transferred directly from the MRI host computer to another computer in a remote location rather than using a removable medium (e.g., film or CD-ROM). The interconnections between the two computers are normally one of two types. An ethernet connection is a physical cable. This is typically used for connecting computers within a building or small group of buildings; for example, between two scanners or between the scanner and a viewing station. Alternately, the connection may be through a dial-up or cable modem, using standard or high-speed telecommunication. This is used for long-distance transmission of images and data. In many facilities, both types of connections are used. Most computer networks use a communications protocol known as transmission control protocol/internet protocol (TCP/IP) for the actual data transfer between computers. Each computer on the network has a unique identification number assigned to it by the network administrator, known as the IP address. This is a series of four numbers, each less than 255, (written as, e.g., 149.123. 65.9), that act as a street address for the computer on the

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 81

Magnetic Resonance Imaging Principles and Applications

network. There will also be one computer on the network known as a router, which facilitates the data transfer. Data transfer is initiated from one computer (e.g., MRI host computer) with the IP address of the destination computer (e.g., viewing station). The data are initially sent to the router, which directs the data to the destination computer based on the IP address. Both the initiating and the destination computers must have each other’s IP address to transfer data in both directions. Although the usage of TCP/IP provides a means for transferring data between computers, there must be a common format for writing the data if it is to be interpreted correctly. This is of particular importance if the two computers use software from different manufacturers, because each manufacturer uses its own proprietary format for data storage. One format that is becoming the industry standard for image and medical data transfer to facilitate such transfer is known as Digital Imaging and Communications in Medicine (DICOM). It is the result of a joint committee of the American College of Radiology and the National Electrical Manufacturers Association (ACR-NEMA). The DICOM standard provides a framework that allows equipment (scanners, digital cameras, viewing stations) from different manufacturers to accept and process data accurately. Images written using this standard have the basic measurement information stored so that any vendor can read and properly display the images with the correct anatomic labeling and basic measurement parameters. Manufacturers who subscribe to the DICOM standard have available a DICOM Conformance Statement, which provides details on how the standard has been implemented. Programmers writing software that use DICOMformatted images should consult the Conformance Statement for the particular manufacturer to ensure proper interpretation of the image header variables. Although a discussion of the complete DICOM standard is beyond the scope of the current discussion, two aspects of it are likely to be encountered with current MRI systems. The first is the format in which images are stored. Images stored in the DICOM format are written with the measurement parameters and other information organized in a fashion as specified in the DICOM Conformance Statement. This allows image display and analysis programs to be written without knowledge of the manufacturer’s proprietary methods. Programs for reading DICOMformat images are available from several companies, with some programs available at no cost and others available for a fee. The second aspect of DICOM that is frequently encountered is in the nature of the data transfer between computers. The DICOM standard has features that control the communication relationship between systems. This is critical to ensure confidentiality of patient information. Two levels of connectivity are commonly used. The first is known as the DICOM Store Service Class (commonly known as Send/Receive). This allows one computer system

81

(e.g., scanner) to send images to another system or to receive images from another system. This is normally used to connect scanners of different modalities (i.e., MR to CT). The other level of connectivity is known as DICOM Query Service Class (commonly known as Query/Retrieve). This allows a remote computer to query the image database on a scanner and retrieve the images without operator intervention. This is the normal connectivity between two scanners of the same modality or between a scanner and an archive server. This connectivity is controlled by the Application Entity Title (AET), and both computers must have matching AETs for the transfer to be successful. The connectivity relationships are assigned by the network administrator during system installation or configuration based on the preferences and policies of the facility.

Magnet System The magnet is the basic component of an MRI scanner. Magnets are available in a variety of field strengths, shapes, and materials. All magnet field strengths are measured in units of Tesla or Gauss (1 Tesla
10,000 Gauss). Magnets are usually categorized as low-, medium-, or high-field systems. Low-field magnets have main field strengths less than 0.5 T. Medium-field systems have main magnetic fields between 0.5 and 1.0 T, high-field systems have fields between 1.0 T and 1.5 T, and ultrahigh field systems have fields of 3.0 T or greater. Magnets are also characterized by the metal used in their composition. Permanent magnets are manufactured from metal that remains magnetic for extremely long periods of time (years). They can be solenoidal (tube shaped) or have a more open design. Permanent magnets have minimum maintenance costs because the field is always present. However, care must be taken to keep ferrous material away from the magnet. Such material will be attracted forcefully into the magnet and the magnetic field cannot be eliminated to allow its extraction. Permanent magnets also have their mass concentrated over a small area. Additional structural support of the scan room may be necessary in some cases. The other types of magnet are electromagnets in which the flow of electrical current through wire coils produces the magnetic field. Traditional electromagnets are made of copper wire wound in loops of various shapes. A power supply provides a constant current source. The magnetic field is present as long as current flows through the magnet windings. Copper wire–based electromagnets are low-field systems and may also be solenoidal or open-type design. The most common type of magnets are solenoidal superconducting electromagnets using niobium-titanium alloy wire immersed in liquid helium as the magnet wire. This alloy has no resistance to the flow of electrical current below a temperature of 20 K. The magnet cryostat, which contains the liquid helium, may be a double-dewar design with a liquid nitrogen container surrounding the helium

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 82

82

Chapter 2

container, or a helium-only design with a refrigeration system to reduce the helium boiloff to a minimum. Refrigeration systems used with current magnets allow helium replenishment rates of once per year. The primary consideration in magnet quality is the homogeneity or uniformity of the magnetic field. One factor affecting magnet homogeneity is the magnet design. Largebore solenoidal magnets generally have the best homogeneity over the largest volume. Short-bore magnets tend to have smaller regions of good homogeneity because of the lower number of magnet windings used. Open-design magnets also have fewer regions of good homogeneity. Magnetic field homogeneity is usually expressed in ppm relative to the main field over a certain distance. High homogeneity means the magnetic field changes very little over the specified region or volume. The protons in this region resonate at the same frequency in a coherent manner and thus induce the maximum possible signal. Great effort is taken during magnet manufacturing and installation to ensure the best homogeneity possible. However, manufacturing imperfections or site problems (e.g., nearby steel posts, asymmetrical metal arrangements) may produce significant field distortions. To analyze and compensate for this, the distortions are characterized by the mathematical shape of the field corrections required as a function of distance away from the magnet center. This classification is referred to by the order of the field or shim correction used. First-order, or linear, corrections in each direction are achieved using the imaging gradient coils described below. Second- or higher-order corrections are nonlinear in nature, and most MRI systems use a coil known as a shim coil to correct for them. The design of the shim coil may be passive, in that it holds pieces of metal (shim plates) or small magnets that correct the field distortions, or active, in that there are loops of wire through which current passes to correct the field distortions. In some systems, both types of shim correction may be used. Passive shimming is generally performed at the time of magnet installation as a one-time event. Active shimming (also called electrical shimming) is usually performed on a regular basis during system maintenance. Field homogeneity is an important factor to consider when evaluating an MRI system because inadequate homogeneity can cause problems with fat saturation or even general imaging. Certain precautions should be exercised around all magnets, regardless of field strength. The examination room in which the magnet is located should have restricted access. Any metal near the magnet should be nonmagnetic. Metal items such as stethoscopes or pens may be attracted to the magnet, causing possible injury. Breathing gases used for sedated patients should be either built into the wall or supplied from nonferrous tanks. Electrical equipment must be protected or shielded from the magnetic field in order to function properly. Patients with surgical implants or metal fragments in their bodies as a result of trauma or occupation (e.g., sheet metal workers)

should be scanned only if there is no risk to the patient should the implant or fragment move during the procedure. Patients with magnetic pacemakers or ferromagnetic intracranial aneurysm clips should not be scanned under any circumstances due to the risk of patient injury. The magnetic properties of various medical implants have been extensively described by Shellock (6). It is also important to realize that the magnetic field of all magnets extends in all directions away from the center of the field. The amount of fringe magnetic field (the portion outside the magnet housing) is a very important consideration in siting an MRI system. The fringe field is greatest near the magnet in the z direction and decreases with increasing distance away from the magnet. The fringe field is also larger for higher-field magnets. A low-field magnet has a very small fringe field, making it easier to use standard patient monitoring equipment. High-field systems are often manufactured with magnetic shielding of some variety to reduce the fringe field. This shielding may surround the magnet (passive shielding), be generated by a second set of superconducting magnet windings surrounding the main field (active shielding), or be built into the wall (room shielding). Two distances are of concern regarding the fringe field. The 0.5-mT (5-G) distance is considered the minimum safe distance for persons with pacemakers. This distance prevents interference of the pacemaker operation by the magnetic field. The 0.1-mT (1-G) distance, the nominal distance for other equipment that uses video monitors, minimizes distortion of the image on the monitor by the magnetic field. The actual distances are installation and equipment specific. Contact the manufacturer regarding individual situations.

Gradient System As mentioned in the section on imaging, small linear distortions to B0, known as gradient fields or gradients, are used to localize the tissue signals. Three gradients are used, one each in the x, y, and z directions, to produce the orthogonal field variations required for imaging. They are each generated by the flow of electrical current through separate loops of wire mounted into a single form known as the gradient coil. Variations in gradient amplitude are produced by changes in the amount or direction of the current flow through the coil. One of the major criteria for evaluation of an MRI scanner is the capabilities of the gradient system. There are four aspects that are important in assessing gradient system performance: maximum gradient strength, rise time (or slew rate), duty cycle, and techniques for eddy current compensation. Gradient strength is measured in units of mT/m or G/cm (1 G/cm
10 mT/m, with typical maximum gradient strengths for current state-of-the-art MRI systems being 30–40 mT/m (3.0–4.0 G/cm). These maximum gradient strengths allow thinner slices or smaller FOVs to be obtained without changing other measurement parameters.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 83

Magnetic Resonance Imaging Principles and Applications

Another quantity often used to describe maximum gradient amplitudes is the effective gradient amplitude. This is the instantaneous vector sum of all three gradients when applied during the scan. Care must be made to distinguish between the effective gradient and the actual gradient being applied. For example, for slice selection, orthogonal slices use only one gradient axis, while only doubleoblique slices require all three axes. The response of a gradient coil to the flow of current is not instantaneous. Gradient pulses require a finite time, known as the rise time, to achieve their final value. This rise time is nominally 0.1–0.5 ms, which defines a rate of change, or slew rate, for a gradient pulse. If the desired gradient pulse amplitude is 20 mT/m with a 0.5-ms rise time, the slew rate is 40 mT/m/ms or 40 T/m/s. Gradient rise times and/or slew rates are often used to evaluate the performance of the gradient amplifier or power supply producing the current. High-performance amplifiers allow shorter rise times (faster slew rates) enabling shorter gradient pulse durations and/or interpulse delays within a pulse sequence. As a result, the minimum TE may be reduced for a given technique while maintaining a small FOV. The duty cycle of the gradient amplifier is another important measure of gradient performance. The duty cycle determines how fast the amplifier can respond to the demands of a pulse sequence. Duty cycles of 100% at the maximum gradient amplitude are typical for state-of-theart gradient amplifiers for normal imaging sequences. Large duty cycles allow high-amplitude gradient pulses to be used with very short interpulse delays. Low duty cycles mean that the TE times will be longer for the scan, to allow the gradient amplifiers to return to a standard operating condition. Another complication of gradient pulses is eddy currents. Eddy currents are electric fields produced in a conductive medium by a time-varying magnetic field. In MRI systems, eddy currents are typically induced by the ramping gradient pulse in the body coil located inside the gradient coil and the cryoshield (the innermost portion of the magnet cryostat) outside the coil. These currents generate a magnetic field that opposes and distorts the original gradient pulse. In addition, once the gradient plateau is reached and the ramp is stopped, the eddy currents begin to reduce. The net result is that the distorting field changes with time as the eddy currents decay. Therefore, the magnetic field homogeneity and the corresponding frequencies change with time as well. Correction of these eddy current–induced distortions is known as eddy current compensation. Two approaches are commonly used for compensation. One method predistorts the gradient pulse so that the field variation inside the magnet is the desired one. This predistortion may be done via hardware or software. A second approach uses a second set of coil windings surrounding the main gradient coil. This approach is called an actively shielded gradient coil, analogous to the actively shielded magnet described previously.

83

The current flow through the shield coil reduces the eddy currents induced within structures outside the coil. Typical state-of-the-art scanners use both methods of eddy current compensation.

Radiofrequency System The rf transmitter system is responsible for generating and broadcasting the rf pulses used to excite the protons. The rf transmitter contains four main components: a frequency synthesizer, the digital envelope of rf frequencies, a highpower amplifier, and a coil or antenna. Each rf pulse has both a frequency and a phase defined for it. These features are determined by the combination of frequency and phase from the frequency synthesizer and the rf envelope defining the pulse shape. The frequency synthesizer produces the center, or carrier, frequency for the rf pulse. It also provides the master clock for the measurement hardware during the scan. The frequency synthesizer also controls the relative phase of the rf pulse. Many pulse sequences alternate the phase of the excitation pulse for each measurement by 180 degrees to help reduce stimulated echo artifacts caused by pulse imperfections. Spin echo sequences also typically have the refocusing rf pulses shifted in phase 90 degrees relative to the excitation pulse (known as a Carr-Purcell-MeiboomGill or CPMG technique). This phase variation may be done through modulation of the rf envelope or of the carrier frequency. More sophisticated synthesizers allow phase changes of 1-degree to 2-degree increments. This finer control also allows for coherence spoiling through incremental phase change of the transmitter, a process known as rf spoiling (see section on artifacts). The rf envelope is stored as a discrete envelope or function containing a range, or bandwidth, of frequencies. It is mixed with the carrier frequency prior to amplification to produce an amplitude- or phase-modulated pulse centered at the desired frequency. For some scanners, the final frequency is produced exclusively by the frequency synthesizer, while for other scanners, the rf envelope is modulated to incorporate a frequency offset into the pulse. In either case, the final frequency is determined based on Eq. 4 and generated as a phase-coherent signal by the synthesizer. The rf power amplifier is responsible for producing sufficient power from the frequency synthesizer signal to excite the protons. The amplifier may be solid-state or a tube type. Typical rf amplifiers for MR scanners are rated at 2–35 kW of output power. The actual amount of power required from the amplifier to rotate the protons from equilibrium depends on the field strength, coil transmission efficiency, transmitter pulse duration, and desired excitation angle. The final component of the rf system is the transmitter coil. All MR measurements require a transmitter coil, or antenna, to broadcast the rf pulses. Although transmitter coils can be any size and shape, the one requirement that

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 84

84

Chapter 2

must be met is that they generate an effective B1field perpendicular to B0. Another feature of most transmitter coils is that they can produce uniform rf excitation over a desired area; that is, a volume can be defined within the coil where all protons experience the same amount of rf energy. Solenoidal MR systems use either a saddle or birdcage coil design, which produces uniform rf excitation even though the coil opening is parallel to B0. These coils are often adjusted, or tuned, to the patient to achieve the maximum efficiency in rf transmission. Two types of coil polarity are used: linear polarized (LP) and circularly polarized (CP), also called quadrature. In an LP system, a single-coil system is present and the rf pulse is broadcast as a plane wave. A plane wave broadcast at a frequency ω TR has two circularly rotating components, rotating in opposite directions at the same frequency, ω TR. For MR, only the component rotating in the same direction as the protons (in-phase) induces resonance absorption. The other component (out-of-phase) is absorbed by the patient as heat. In a CP transmitter system, two coils are present, one rotated 90 degrees from the other. Phase-shifted rf pulses are broadcast through each coil. The out-of-phase components cancel each other out while the in-phase components add coherently. The patient absorbs only the energy from the in-phase components from each coil. A 40% higher efficiency from the transmitter system is achieved for a CP system relative to an equivalent LP system for the same proton rotation. Although MR is considered a relatively safe imaging technique, the absorbed rf power generates heat inside the patient. Manufacturers are required by the governing regulatory organizations [e.g., the U.S. Food and Drug Administration (FDA), International Electrotechnical Commission (IEC)] to monitor the rf power absorbed by the patient so that excessive patient heating does not occur over both the excited tissue volume (localized) and the entire patient. To accomplish this, the SAR is monitored. The SAR is measured in Watts of energy per kilogram of patient body weight (W/kg). MRI systems are designed to operate at or below the SAR guidelines, which are set to limit the patient heating to approximately 1°C or less. For low-field scanners, the SAR seldom limits the measurement protocols. For high-field scanners, the SAR limits have a significant effect on the number of slices or saturation pulses that can be applied to the patient within a scan.

Data Acquisition System The data acquisition system is responsible for measuring the signals from the protons and digitizing them for later postprocessing. All MRI systems use a coil to detect the induced voltage from the protons following an rf pulse. The coil is tuned to the particular frequency of the returning signal. This coil may be the same one used to broadcast the rf pulse, or it may be a dedicated receiver coil. The exact shape and size of the coil are manufacturer specific,

but its effective field must be perpendicular to B0. The sensitivity of the coil depends on its size, with smaller coils being more sensitive than larger coils. Also, the amount of tissue within the sensitive volume of the coil, known as the filling factor, affects the sensitivity. For large-volume studies, such as body or head imaging, the transmitter coil often serves as the receiver coil. For smaller-volume studies, receive-only surface coils are usually used. These coils are small, usually ring shaped, have high sensitivity but limited penetration, and are used to examine anatomy near the surface of the patient’s body. Phased array coils use two or more smaller surface coils to cover a larger area. The small coils are configured so that there is minimal interference between them. This arrangement provides the sensitivity of the small coil but with the anatomical coverage of the larger coil. The signals produced by the protons are usually 1 to 2 microvolts in amplitude and in the MHz frequency range. To process them, amplification is required, which is usually performed in several stages. The initial amplification is performed using a low-noise, high-gain preamplifier located inside the magnet room or built into the coil itself. This signal is further amplified, then demodulated to a lower frequency. Further digital amplification is performed by the receiver module. Two types of receivers are used in MRI systems. Analog receivers amplify then demodulate the measured signal relative to the input frequency from the frequency synthesizer to produce a quadrature signal (real and imaginary) with frequencies of 1,000 to 250,000 Hz (audio frequency). The signals are filtered with band-pass filters, then digitized using ADCs. The ADCs digitize each analog signal at a rate determined by the sampling time and number of data points specified by the user. Typical ADCs can digitize a 10-V signal into 16 bits of information at a rate of 0.5 to 1,000 s per data point. Digital receivers demodulate the signal to an intermediate frequency, and then digitize it using an ADC. The final signal amplification, application of a low-pass filter, demodulation to an audio range, and quadrature formulation are done mathematically on the digital signal. Digital receivers allow for better signal fidelity of the final signal than analog receivers. The digitized data are stored on a hard disk or in computer memory for later Fourier transformation. Phased-array coils typically have a separate preamplifier and ADC for each coil in the array. Although not formally part of the data acquisition system hardware, an important component of an MRI scanner is rf shielding of the scan room. The weak MR signals must be detected in the presence of background rf signals from local radio and television stations. To filter this extraneous noise, MRI scanners are normally enclosed in a copper or stainless steel shield known as a Faraday shield. Maintaining the integrity of this Faraday shield is very important to minimize noise contamination of the final images.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 85

Magnetic Resonance Imaging Principles and Applications

85

CONTRAST AGENTS One of the strengths of MRI is the significant amount of intrinsic contrast between tissues. This contrast is based on differences in signal intensity between adjacent pixels in the image. It is a result primarily of differences in the T1 and/or T2 relaxation times of the tissues under observation, accentuated by the chosen TR and TE. Pathologic tissue may or may not have significant differences in T1 or T2 from the surrounding normal tissue. For this reason, there may be little signal difference between normal and pathologic tissue in spite of the inherent high contrast in the images. This signal difference can be increased through the administration of a contrast agent. Contrast agents for MRI have several advantages over those used for CT. CT agents are direct agents in that they contain an atom (iodine, barium) that attenuates or scatters the incident x-ray beam differently from the surrounding tissue. This scattering permits direct visualization of the agent itself regardless of its location. Most MRI contrast agents are indirect agents in that they are never visualized directly in the image, but affect the relaxation times of the water protons in the nearby tissue. The concentration and dosage for MRI agents are also significantly lower than those for CT agents, which in part explains the lower occurrence of adverse reactions with MRI agents. They are normally excreted through the renal system within 3 to 4 days, though the excretion half-time varies substantially from agent to agent. Contrast agents are usually categorized as T1 or T2 agents based on their primary effect of shortening the T1 or T2 relaxation times, respectively. However, these agents also shorten the other relaxation time to a lesser degree. The amount of reduction depends on the concentration of the agent: when concentrated, T1 agents shorten T2 or T2* times and when dilute, T2 agents reduce T1 times. Contrast agents may also be grouped into intravenous and oral agents depending on the route of administration. The following is a brief discussion of MR contrast agents with emphasis on those in current clinical use.

Intravenous Agents T1 Relaxation Agents Most intravenous contrast agents currently in clinical use are T1 relaxation agents. A variety of formulations are available, all of which consist of one or more paramagnetic metal ions that contain one or more unpaired electrons. The metal ions are either bound into a chelate complex or contained within a macromolecule. The relaxation efficiency per molecule depends on several factors, including the nature of the metal ion and the size and structure of the metal-chelate complex. The primary mode of operation for T1 contrast agents is as a relaxation sink for the water protons. As mentioned previously, T1 relaxation depends on the “lattice” receiving the energy that the protons have absorbed from the rf excitation pulse. This energy

H

H O

H

H O

Gd

Figure 2-76 Exchange of water molecules in a coordination sphere of gadolinium (Gd)-chelated contrast agent. The chelate molecule causes steric hindrance (crowding) around the Gd atom, restricting its access. An excited water molecule is small enough to reach the inner sphere and transfer its energy to the gadolinium ion, then leaves the complex unexcited as another molecule replaces it.

transfer occurs most efficiently when the protons are in the innermost layer of atoms surrounding the metal ion, known as the coordination sphere. Because the chelate molecules are relatively large and have many bonds with the metal ion, there is limited free space in the coordination sphere, which prevents the protons in large molecules, such as fat, from getting sufficiently close to the metal ion for efficient energy transfer. The tissue water is able to diffuse into the coordination sphere of the metal ion and give up its energy, then exchange with the bulk tissue water, enabling additional water molecules to enter the coordination sphere. This diffusion/exchange happens very rapidly (~106 times per second) so that the bulk tissue water is relaxed when the subsequent excitation pulse is applied (Fig. 2-76). The result is that the tissue water near the contrast agent has a larger net magnetization, M, than water in the neighboring tissue and will contribute more signal in a T1-weighted image (see Fig. 2-74). In addition, the rapid exchange process enables many water molecules to be affected by a single chelate complex, which allows low concentrations of contrast agent to be used in clinical studies. MR contrast agents were originally developed for use in neuroimaging to differentiate between normal and neoplastic tissues, but their usage has extended to other examinations including routine and angiographic imaging in the abdomen and thorax (Fig. 2-77). Intravenous T1 contrast agents are available in both ionic and nonionic formulations. The most common agents use gadolinium as the metal ion and either diethylenetriaminepentaacetate (DTPA) (ionic; Magnevist; Schering AG, Berlin, Germany), diethylenetriaminepentaacetate bismethylamide (DTPA-BMA) (nonionic; Omniscan; GE Healthcare, Chalfont St. Giles, United Kingdom), or 10-(2-hydroxypropyl)1,4,7,10-tetraazacyclododecane-1,4,7-triacetate (HP-DO3A) (nonionic; Prohance; Bracco Diagnostic, Princeton, NJ) as

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 86

86

Chapter 2

Figure 2-77 Magnetic resonance angiography of renal arteries acquired following administration of a gadolinium-chelated contrast agent.

the chelate molecule. The agents currently licensed for use in the United States are nonspecific extracellular agents, in that the agents diffuse rapidly from the vascular space into the interstitial space of the tissue. The major elimination pathway of these agents is through glomerular filtration and renal excretion. The half-life is typically on the order of 90 minutes, with virtually complete elimination of these agents within 24 hours. The safety profile for these agents is excellent, with few reported side effects and rare cases of major side effects. These agents are therefore frequently employed in patients with a known history of allergies to iodinated contrast agents. In addition, these agents are considered to be extremely safe for administration to patients with compromised renal function, due to both their intrinsic safety and the relatively small volume of the agent administered for most studies (typically 10 to 20 mL). On the other hand, the adverse effect of iodinated contrast in patients with renal failure is a well-known problem. Therefore, patients with renal failure or elevated levels of serum creatinine are more likely to tolerate an MRI examination with administration of a gadoliniumchelated contrast agent rather than a CT examination using iodinated contrast media. Other formulations of gadolinium-chelated contrast agents have both hydrophilic and hydrophobic properties. Two examples of these use benzoxypropionic tetraacetate (BOPTA) and ethoxybenzyl diethylenetriaminepentaacetate (EOB-DTPA) as the ligand. The presence of a hydrophobic ligand enables hepatocyte uptake of these agents and elimination by the biliary system, rather than strictly through renal excretion. Approximately 10% of GdBOPTA (Multihance; Bracco Diagnostic, Princeton, New Jersey) is eliminated by the hepatobiliary system and 90% by renal excretion, whereas 50% of Gd-EOB-DTPA (Eovist; Schering AG, Berlin, Germany) is eliminated by the hepatobiliary system and 50% by renal excretion. The dual pathways of elimination for these agents enable postcontrast imaging emphasizing both tissue perfusion (immediately following administration) and hepatocellular uptake

and bile duct elimination (typically between 10 and 60 minutes). These agents are currently not licensed for clinical use in the United States; however, the combined effect of their ease of use, safety, and dual-phase behavior suggests that they have great potential as liver contrast agents. Another metal ion that has been formulated into a T1 contrast agent for liver imaging is manganese (Mn). Manganese ion chelated to N,N’-dipyridoxal ethylenediamineN,N’-diacetate 5,5’ bis(phosphate) (DPDP) (Mangofodipir; Teslascan; Amersham, New York, New York) is a T1 contrast agent that is hepatocyte selective but is also absorbed by other organs and tissues, including the pancreas and the adrenal and renal cortexes. Approximately 70% of the agent is eliminated by the hepatobiliary system and 30% by the urinary system. Mn-DPDP is administered via a slow intravenous injection over 1–2 minutes. This agent has a long imaging window with maximal hepatic parenchymal enhancement lasting from 15 minutes to 4 hours. The major clinical indication for the use of MnDPDP is in the determination of the extent of liver metastases in patients under consideration for surgical resection. Mn-DPDP may also be used to characterize whether liver lesions contain hepatocytes and to improve delineation of pancreatic tumors. Adverse reactions are relatively uncommon with this agent. Other T1 contrast agents have been developed using gadolinium ions chelated to macromolecules or polymers. These agents remain within the blood stream rather than diffuse into the tissue. Their use as MR angiographic agents and as agents to assess the integrity of capillary basement membrane is currently under investigation.

T2 Relaxation Agents The other class of intravenous contrast agents is T2 relaxation agents. They are typically macromolecules containing several iron atoms that collectively form a superparamagnetic center. The large magnetic susceptibility of the macromolecule distorts the local magnetic field in the vicinity of the agent, causing the nearby water protons to dephase more rapidly than the surrounding tissue. This condition results in significant signal loss in T2-weighted spin echo or gradient echo images. The most common T2 contrast agents are based on superparamagnetic iron oxide (SPIO) particulate molecules, also known as ferumoxides. These agents are selectively absorbed by the reticuloendothelial cells located in the liver, spleen, and bone marrow. Normal tissue in these organ systems takes up the agent and has a low signal on T2- or T2*-weighted images. Lesions that do not contain reticuloendothelial cells in appreciable numbers do not take up the agent and remain unaffected, and therefore have a relatively high signal. Currently, the only T2 contrast agent licensed for clinical use in the United States is magnetite-dextran (Feridex; Schering AG, Berlin, Germany). It is usually administered as a slow drip infusion over 30 minutes. A further delay of 30 minutes prior to imaging allows for maximal uptake of

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 87

Magnetic Resonance Imaging Principles and Applications

the agent by reticuloendothelial cells. Contrast enhancement can be observed from 30 minutes to 4 hours following infusion. Magnetite-dextran has a good safety profile, but has a potential side effect of acute back pain developing during infusion, occurring in less than 3% of patients. This side effect is usually self-limiting and disappears when the infusion is stopped or slowed. The most important clinical indication for magnetite-dextran is in the determination of the extent of liver metastases in patients under consideration for surgical resection. This agent can also distinguish between hepatic origin tumors that contain reticuloendothelial cells and tumors that do not. Magnetite/maghemite-carboxydextran (Resovist; Schering AG, Berlin, Germany) is another formulation of a ferumoxide agent that can be administered in a small dose by bolus injection. The advantage of this agent is ease of administration and the lack of back pain as a side effect. It is an ultrasmall particulate iron oxide (USPIO) that also increases T1 contrast in dynamic gradient echo scanning immediately following administration, enabling one agent to affect both T1- and T2-weighted scans. Resovist is currently under consideration for approval by the U.S. FDA. SPIO agents formulated to a smaller particle size than that used for liver imaging are under investigation as contrast agents for the examination of lymph nodes. Normal or hyperplastic lymph nodes take up the agent and lose signal on T2-weighted images, whereas malignant lymph nodes do not take up the agent and therefore have a relatively high signal. Iron oxide particles have also been used to label monoclonal antibodies targeted to specific tissue receptor sites. Asioglycan protein receptor contrast agents are one example that is under development.

Oral Agents Oral MRI contrast agents are typically nonspecific in nature. They are most often used in abdominal and pelvic studies to provide reliable differentiation of bowel from adjacent structures and to provide better delineation of bowel wall processes. Oral agents may be categorized as positive or negative agents. Positive agents increase the overall signal intensity within the image, generally by shortening the T1 or T2 relaxation times of tissue water. These agents are generally solutions of paramagnetic metal ions or metal-chelate complexes. Many of these agents are present in naturally occurring products (manganese in green tea and blueberry juice) or in over-the-counter medications (ferric ammonium citrate; Geritol; Beecham, Bristol, Tennessee). Other agents have been specifically formulated for use with MRI such as manganese chloride (Lumenhance; Bracco Diagnostic, Princeton, New Jersey) and gadolinium-DTPA (Magnevist Enteral; Schering AG, Berlin, Germany) as T1 relaxation agents and ferric ammonium citrate (OMR; Oncomembrane, Seattle, Washington) as a T2 relaxation agent. Positive agents can provide excellent delineation of the bowel, but the increased signal may

87

induce greater artifacts as a result of respiratory motion or peristalsis. The other type of oral contrast agent is a negative agent. Negative agents eliminate tissue signal from the area of interest. Two approaches are used for negative agents. One method reduces the T2 relaxation times using suspensions of ferumoxide particles. The particles may be suspended in aqueous solution (Ferumoxsil; Advanced Magnetics, Cambridge, Massachusetts) or adsorbed onto a polymer. The other approach uses an agent that contains no protons, and therefore produces no visible MR signal. Barium sulfate, clay, and air have been used for intraluminal studies. One significant problem with many negative-contrast agents, particularly the iron-based agents, is that they increase the local magnetic susceptibility, which may induce significant dephasing artifacts at high field strengths.

CLINICAL APPLICATIONS In selecting pulse sequences and measurement parameters for a specific application, MRI allows the user tremendous flexibility to produce variations in contrast between normal and diseased tissue. This flexibility is available when imaging both stationary tissue as well as flowing blood. For example, the use of both bright-blood and dark-blood MRA techniques permits more accurate assessment of vascular patency and intravascular mass lesions. A typical patient examination acquires sets of images with multiple types of contrast (proton density, T1, T2) and multiple slice orientations (transverse, sagittal, coronal, oblique), providing the clinician with more complete information on the nature of the tissue under observation and increasing the likelihood of lesion detection. It is important that the MR examinations be tailored to the organ or organs under investigation, the type of disease process, and the individual patient. The MR physician may choose to provide detailed measurement parameters for each examination, or have a predetermined regimen of scans that are to be performed by a technologist. Establishing fixed measurement protocols ensures the efficient operation of the MR scanner and that reliable, reproducible imaging examinations are performed.

General Principles of Clinical Magnetic Resonance Imaging The fundamental goals of diagnostic MRI of the body are: (a) accurate and reproducible image quality; (b) good visualization of disease processes; and (c) comprehensive imaging information for the area under observation. Although the ideal achievement of all three goals may not be practical, the proper choice of pulse sequences and measurement parameters can provide adequate results within a clinically acceptable scan time. In particular, the ability to visualize disease relative to normal tissue can be dramatically affected by the measurement protocol. One common

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 88

88

Chapter 2

approach to enhance the contrast between abnormal and background tissue is to make the signal of one of the tissues significantly different from the other one. For example, variation of the signal from fat hydrogen through the use of fat-unsuppressed and fat-suppressed techniques aids in the detection of lymph nodes. On T1-weighted images, lymph nodes have a low signal and therefore are conspicuous against a background of high-signal fat, whereas on T2-weighted images, lymph nodes are relatively bright, and their conspicuity is improved by decreasing the signal of background fat using fat-suppressing techniques. Following administration of a gadolinium-chelated contrast agent, lymph nodes have significantly shorter T1 values and produce a high signal on T1-weighted images. The use of fat suppression is helpful to reduce the competing high signal of fat. The combination of gadolinium-chelated contrast agents to increase the signal from diseased tissue and fat suppression to decrease the signal from background fat is widely used in various organ systems, including the orbits, the bony skeleton, soft tissue of the extremities, and the breast. An additional consideration is in the choice of pulse sequence, in that there may be differences in signal from flowing tissue or dramatic differences in image quality. For example, fat-suppressed spoiled gradient echo may be preferable to fat-suppressed spin echo when concomitant evaluation of patency of vessels is desired, as in imaging vascular grafts for patency or infection, or for imaging extremities to assess soft tissue infection or vascular thrombosis. This is because spoiled gradient echo images, when acquired within 2 minutes following administration of gadolinium-chelated contrast media, display patent vessels with high signal intensity and occluded vessels with low signal intensity, while spin echo images may have a signal void from both patent and thrombosed vessels. Flowing tissue is commonly seen as a signal void on spin echo images because of the dephasing of moving spins during the gradient pulses, whereas blood clots produce a low signal owing to the presence of a fibrinous clot and greater T2* effects from blood breakdown products. It is important when defining protocols for MRI studies to obtain a sufficient variety of sequences to provide comprehensive information, while at the same time not to be too redundant and generate excessively long exams. For example, our approach for imaging of the abdomen has been to employ a variety of short-duration T1- and T2weighted sequences with the majority of them performed in the transverse plane, but also with at least one set of images in a plane obtained orthogonal to the transverse plane. The particular choice of planes for a measurement is dictated by the area of anatomy under observation.

Examination Design Considerations The initial studies of MRI used primarily transverse T1and T2-weighted spin echo techniques. As the modality

has matured, it has become clear that imaging strategies must be modified beyond this elementary approach. There are several reasons for this: 1. Imaging of organs in the plane of the best anatomical display. Imaging of the female pelvis requires the use of both sagittal and transverse images to best demonstrate the anatomical structures. Similarly, the evaluation of large masses in the region of the upper poles of the kidneys is facilitated by sagittal as well as transverse images. On the other hand, coronal images are useful for visualization of the left lobe of the liver. 2. Compensation for the most severe artifacts generated by various organ systems. Abdominal and thoracic imaging are severely compromised by artifacts from respiratory motion. Imaging protocols that employ breathhold spoiled gradient echo sequences (e.g., fast low-angle shot [FLASH], spoiled gradient recalled acquisition in the steady state [GRASS]) can produce images with substantial T1 weighting while minimizing respiratory artifacts. The use of cardiac gating or very short TR/TE sequences is valuable when imaging the thorax in order to minimize phase artifacts from cardiac motion. 3. Increased spatial resolution and/or S/N. Imaging of small anatomical regions such as the breast requires specialized surface coils to maximize both S/N and spatial resolution. 4. The use of contrast agents. Imaging of nonorgan-deforming focal lesions in the liver, spleen, and pancreas using intravenous gadolinium-chelated complexes requires rapid, breathhold imaging techniques to capture the capillary phase of lesion enhancement. This approach necessitates the use of spoiled gradient echo techniques (e.g., FLASH or spoiled GRASS) with a temporal resolution of less than 20 seconds. In many instances, the visualization of contrast enhancement in T1-weighted images may be improved through the use of fat suppression to remove the competing high signal from fat.

Protocol Considerations for Anatomical Regions The following considerations and recommendations are offered for imaging various organ systems. They are intended to indicate general guidelines for use in developing imaging examinations. Exact sequence protocols are not provided because different manufacturers provide different imaging capabilities and functions.

Thorax Cardiac triggering is useful to minimize motion artifacts from the heart. Transverse T1-weighted spin echo images provide good anatomical evaluation of the mediastinum and chest wall. T2-weighted transverse images are helpful, particularly in the evaluation of chest wall or mediastinal involvement with cancer. T1-weighted images acquired fol-

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 89

Magnetic Resonance Imaging Principles and Applications

lowing administration of gadolinium-chelated contrast media provide similar and complementary information. Fat suppression is a useful adjunct when gadoliniumchelated contrast agents are employed. The reduced signal from fat improves the visualization of abnormal tissue enhancement, which is helpful for delineating the presence of chest wall invasion by malignant or infectious processes. Lesions in the lung apex and occasionally in the lung base require additional coronal or sagittal views. Breathhold imaging following administration of gadolinium-chelated contrast media improves visualization of small peripheral lung lesions; metastases can be reliably seen at a diameter of 5 mm using this approach. A 3D gradient echo sequence with a very short TE (1–2 ms) is preferred for imaging lung parenchyma because the technique minimizes phase artifacts from cardiac motion. For most examinations, imaging between 2 and 5 minutes following contrast administration provides a good balance between contrast enhancement of lung masses and diminished enhancement of the blood pool.

Heart and Great Vessels Transverse cardiac-triggered T1-weighted spin echo or segmented gradient echo sequences are useful in the evaluation of cardiac anatomy. When possible, breath-hold scanning or the use of navigator echoes is preferred to minimize artifacts from respiratory motion. High spatial resolution is frequently needed, particularly in assessing congenital heart disease; therefore, a slice thickness of less than 5 mm is recommended. Coronal and sagittal cardiactriggered T1-weighted images may provide additional information in many instances. They are essential in the evaluation of congenital heart disease by providing anatomical information regarding vessels, airways, and cardiac chambers. The left anterior oblique sagittal plane is an important view for the evaluation of the thoracic aorta. Signal from flowing blood within the cardiac chambers is often heterogeneous as a result of changes in flow direction and velocity. For T1-weighted sequences, it is helpful to minimize the blood signal, both to reduce flow artifacts and to delineate vessels using dark blood techniques. Application of superior and inferior presaturation pulses or gradient dephasing may be required. Flow compensation should be avoided because it results in an increase in the signal from flowing blood. Alternatively, techniques that result in a high signal from flowing blood (bright blood techniques) are often helpful for the evaluation of chamber dynamics. Multiphasic techniques (cine MR) are particularly useful in the evaluation of wall motion and thickening, valvular disease, and shunts. Useful image orientations include the sagittal plane for evaluating the pulmonary valve, the coronal and left ventricular long axis for evaluating the aortic valve, and the transverse plane for evaluating the tricuspid and mitral valves and for assessing atrial and ventricular septal defects (Fig. 2-78). Single-shot ETSE

89

Figure 2-78 Oblique coronal gradient echo image of the heart using a bright-blood technique. TR, 35 ms; TE, 1.4 ms.

techniques can minimize cardiac motion artifacts and are useful as bright or dark blood techniques (Fig. 2-79). For the evaluation of the aorta and major branches, a widely used technique is MR angiography employing a dynamic 3D gradient echo sequence following gadolinium-chelated contrast media administration.

Liver Both T1- and T2-weighted transverse images are important for evaluating the liver. Combining breathhold and nonbreathhold sequences is often useful since some patients can suspend respiration but cannot breathe regularly, while other patients breathe regularly but cannot hold their breath well. Most current protocols combine breathhold T1- and T2-weighted sequences with breathing-independent T2-weighted sequences. A

Figure 2-79 Oblique transverse echo train spin echo image of the heart using a dark-blood technique. TR, 1,587 ms; effective TE, 67 ms.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 90

90

Chapter 2

Figure 2-80 Transverse T1-weighted spoiled gradient echo liver image. TR, 142 ms; TE, 4 ms; excitation angle, 70 degrees.

Figure 2-82 Coronal single-shot echo train T2-weighted image of the liver and kidneys. Effective TE, 90 ms.

spoiled gradient echo sequence is most often used for T1-weighted imaging (Fig. 2-80); however, T1-weighted spin echo with respiratory compensation may be considered for some patients. For T2-weighted images, breathing-averaged ETSE with fat suppression is a useful technique (Fig. 2-81). Compared with standard T2weighted spin echo techniques, the imaging time is significantly reduced. The addition of fat suppression diminishes respiratory ghosts, removes CSAs, and diminishes the signal of a fatty-infiltrated liver, which permits good visualization of the liver capsular surface and facilitates lesion detection. In many cases, T2weighted single-shot ETSE techniques acquired with and without fat suppression may be sufficient (Fig. 2-82). The use of intravenous contrast agents significantly improves lesion detection over nonenhanced imaging

Figure 2-81 Transverse echo train inversion recovery liver image. TR, 5,000 ms; effective TE, 81 ms; TI, 170 ms; echo train length, 31.

techniques. Currently available contrast agents include gadolinium-chelated complexes and Mn-DPDP for T1 enhancement and ferumoxides for T2 enhancement. When using gadolinium-chelated contrast agents, it is important to image early after contrast (
30 seconds) to maximize the specific enhancement features of various focal hepatic lesions (see Fig. 2-74). Spoiled gradient echo techniques are extremely useful to accomplish this goal. In addition, serial sequence repetition with additional acquisitions at approximately 1 and 5 minutes may be routinely useful to visualize the temporal behavior of contrast uptake. While traditional spoiled gradient echo imaging uses 2D techniques, 3D gradient echo techniques together with fat suppression are continuously improving in image quality and are the method of choice in many facilities for dynamic contrast examinations (Fig. 2-83). Mn-DPDP has a longer imaging window (30 minutes to 4 hours), so it lacks dynamic contrast enhancement ability. Because of this, it is not essential that the patient be able to suspend respiration because dynamic imaging is not performed. High-resolution (512 matrix) spoiled gradient echo is a useful sequence in combination with Mn-DPDP in that it permits detection of smaller focal lesions in conjunction with the T1-shortening effect of the agent. Fat suppression is another useful modification to employ with Mn-DPDP with the fat suppression serving to maximize the conspicuity of contrast enhancement. This may be particularly beneficial for imaging the pancreas. Like Mn-DPDP, magnetite-dextran also has a long imaging window of 30 minutes to 4 hours, and does not require the patient be able to suspend respiration. When using a ferumoxide agent, an important consideration for the choice of sequence is the dose of iron administered. In Europe, a dose of iron is used that is one and

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 91

Magnetic Resonance Imaging Principles and Applications

91

B

A Figure 2-83 Transverse T1-weighted spoiled gradient echo liver images with fat suppression. A: Two-dimensional gradient echo; TR, 242 ms; TE, 4.8 ms; excitation angle, 70 degrees. B: Three-dimensional gradient echo; TR, 4.8 ms; TE, 2.3 ms; excitation angle, 10 degrees.

one half times that of the dose used in North America. This higher dose permits a greater choice of sequences that may be effective, including gradient echo sequences that are fundamentally T1 weighted. Because of this, care must be exercised in extrapolating information from European studies to American studies, particularly in regard to sequence use. Sequences that are useful in conjunction with ferumoxide agents include fat-suppressed ETSE, single-shot ETSE, echo train STIR, and gradient echo sequences with a relatively long TE and relatively low flip angle (e.g., TR
150 ms, TE
9 ms, flip angle
45 degrees). Examining for fatty infiltration requires out-of-phase gradient echo images prior to contrast administration.

Pelvis Transverse images are routinely used in evaluating the pelvis, and sagittal images provide a useful adjunct. T1weighted images are frequently best performed as conventional spin echo or spoiled gradient echo, whereas T2weighted images may be acquired as breathing-averaged ETSE or single-shot ETSE. Thin (5-mm) slices in the sagittal plane are necessary to optimally visualize the uterus and ovaries in the female and the seminal vesicles in the male (Figs. 2-84 to 2-86). Detailed imaging of the pelvis

Other Abdominal Organs T1-weighted breath-hold spoiled gradient echo and T1weighted fat-suppressed spin echo are both useful for controlling respiratory artifacts in the abdomen. As with the liver, combining breath-hold with breathing-independent sequences is advantageous. Single-shot ETSE T2-weighted sequences are particularly effective at demonstrating the bowel and for distinguishing bowel from other entities. Transverse fat-suppressed and conventional spoiled gradient echo images are useful in combination both prior to and following administration of intravenous gadoliniumchelated contrast agents. Following contrast administration, the unsuppressed technique is useful to visualize capillary phase enhancement while the subsequent use of the fat-suppressed technique provides interstitial phase information. For the kidneys and pancreas, dynamic acquisition of fat-suppressed 3D gradient echo images enables excellent organ visualization. Imaging of the adrenal glands requires noncontrast out-of-phase gradient echo images.

Figure 2-84 Coronal spin echo T1-weighted pelvis image. TR, 437 ms; TE, 14 ms.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 92

92

Chapter 2

Spoiled Gradient Echo Imaging parameters include a relatively long TR (
140 ms) and the shortest in-phase echo time (4.5 ms at 1.5 T, 6.67 ms at 1.0 T), one signal average, and an excitation angle of 70 to 90 degrees. These parameters maximize the number of slices, S/N, and T1 weighting, yet minimize artifacts and the measurement time. This spoiled gradient echo sequence is ideal for T1-weighted imaging in the abdomen, for imaging in multiple planes, and for imaging following administration of intravenous gadoliniumchelated contrast media. When thin sections and high spatial resolution are desired and respiratory artifacts are not a problem, T1-weighted spin echo may be preferred. Figure 2-85 Transverse echo train spin echo proton density– weighted image of the pelvis. TR, 4,530 ms; effective TE, 27 ms; echo train length, 7.

also requires high in-plane spatial resolution (512  512 matrices, ETSE, small FOV). Gadolinium-chelated contrast administration is important in the evaluation of uterine and adnexal masses. Fat suppression is an important adjunct to contrast administration when ovarian cancer is clinically suspected.

Recommendations for Specific Sequences and Clinical Situations T1-Weighted Techniques Single Echo Spin Echo For routine T1-weighted imaging, a moderate TR is recommended (400–700 ms at 1.0 T and 1.5 T) with a short TE (20 ms or less). This combination allows acquisition of a sufficient number of slices yet provides acceptable contrast between most tissues. The slice thickness and in-plane spatial resolution can be tailored to the particular anatomical region under examination.

Figure 2-86 Transverse echo train spin echo T2-weighted image of the pelvis. TR, 4,530 ms; effective TE, 137 ms; echo train length, 7.

Out-of-Phase Gradient Echo Imaging parameters should include the shortest possible TE, preferably shorter than the in-phase TE (2.25 ms at 1.5 T, 3.3 ms at 1.0 T). This technique allows easy detection of fat and water in similar proportions within a tissue volume through signal cancellation. For clinical use, the most important applications for this technique are the examination for fatty infiltration of the liver and for benign adrenal adenoma. In both cases, the observation of a signal drop in out-of-phase images compared with in-phase images permits the diagnosis of increased lipid content. For adrenal glands, this technique is virtually pathognomonic for benign disease in which the signal loss is uniform. Fat Saturation This technique is ideal when an expanded range of soft tissue signal intensities and decreased phase artifact from moving tissues containing fat are desired. Improved demonstration of tissue with a high protein content (e.g., normal pancreas) or detection of subacute blood are important uses for T1-weighted fat saturation techniques. When used following administration of gadoliniumchelated contrast media, the removal of the high signal intensity of fat results in easier discrimination of diseased tissue (e.g., diseased bowel or peritoneum, breast cancer, musculoskeletal neoplasms or inflammation). Fat-suppressed spin echo or spoiled gradient echo imaging is also useful in assessing the fat composition in certain masses, such as adrenal myelolipoma, colonic lipoma, or ovarian dermoid cyst. Unlike out-of-phase gradient echo techniques, in which the maximal signal loss occurs in tissues where the fat and water are of equal proportions, fat suppression is most effective when the fat content in the tissue approaches 100%. STIR STIR is not as sensitive to magnetic field homogeneity as fat saturation is and therefore is well suited for imaging small anatomic regions, regions away from the magnet isocenter, and regions where there are large differences in magnetic susceptibility. Images resemble those obtained

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 93

Magnetic Resonance Imaging Principles and Applications

using fat-suppressed T2-weighted spin echo techniques. STIR sequences acquired using TI times for fat and for silicon suppression are useful for the investigation of breast implant rupture. Because long TR times are necessary, they are usually implemented with an echo train to dramatically shorten acquisition time. Performed as a breath-hold scan, this technique has achieved clinical utility in liver imaging.

T2-Weighted Techniques Standard Multiecho Spin Echo Standard multiecho spin echo techniques are used when subtle differences in T2 between normal and diseased tissue are expected. A long TR (greater than 2,000 ms at 1.0 T and 1.5 T) is used to minimize T1 saturation of most tissues, though CSF and other fluids are significantly suppressed in signal due to saturation. A short TE (less than 20 ms) is used to produce proton density–weighted images while a long TE (80 ms or greater) generates T2-weighted images. Gradient motion rephasing is normally used on the long TE image in brain and spine imaging to minimize flow artifacts from bright CSF. Echo Train Spin Echo ETSE techniques are useful when high spatial resolution or decreased imaging time is desired for T2-weighted images (e.g., brain, spine, or pelvic imaging). These sequences are best used when T2 differences between normal and diseased tissues are significant, since they result in an averaging of T2 information. Use of a very long TR (greater than 3,500 ms) allows coverage of anatomic regions and reduces the amount of signal loss from CSF saturation. Caution should be employed when using ETSE techniques for imaging the liver, particularly in the investigation for hepatocellular carcinoma, because the T2 difference between tumor and background liver may be slight. This is normally not a problem if this technique is used in conjunction with spoiled gradient echo immediately following administration of gadolinium-chelated contrast media, as the latter technique is the preferred method for tumor detection and characterization. Fat Saturation Fat-suppressed T2-weighted spin echo is useful for liver or musculoskeletal imaging. It results in a decreased phase artifact from respiratory motion of the abdominal wall and provides an expanded dynamic range of tissue signal intensities and greater conspicuity for focal lesions. It is an excellent technique for imaging capsular-based disease (e.g., hepatosplenic candidiasis or metastases spread through peritoneal seeding) since CSAs are absent. Fat saturation can also be applied to echo train and single-shot echo train sequences. It is essential to use fat saturation when performing T2-weighted ETSE sequences of the liver

93

to attenuate the signal of fat in the setting of fatty liver, because the high signal of fatty liver on nonsuppressed ETSE sequences can mask the presence of liver lesions. The use of fat suppression on T2-weighted sequences also facilitates the detection of lymph nodes, which appear relatively bright against a background of low fat signal.

Sedated or Agitated Patients Optimal MR imaging results require cooperation from patients to minimize motion artifacts. Pads or restraints may be used to restrict patient movement. Patients who are unable to hold their breath because of sedation or decreased consciousness cannot be imaged with breathhold imaging. Spin echo sequences may produce images of good quality in sedated patients, but poor image quality may result from agitated patients. Use of reordered phase encoding or navigator echoes may prove advantageous. For dynamic scanning following administration of gadolinium-chelated contrast media, images can be obtained using slice selective 180-degree inversion pulse prepared snapshot gradient echo sequences (e.g., slice selective TurboFLASH, IR-prepared GRASS) followed by either regular or fat-suppressed T1-weighted spin echo techniques. Patients who are unable to hold their breath or who are slightly agitated require imaging using rapid techniques that are motion insensitive. These include snapshot inversion pulse prepared gradient echo sequences for T1-weighted images and snapshot ETSE for T2-weighted images. These techniques are less sensitive to respiratory motion, and acceptable image quality may be obtained. Imaging studies of sedated patients can produce results of good quality when combining breathing-averaged spin echo or navigator echo-based techniques with single-shot approaches.

REFERENCES 1. Edelman RR, Chen Q. EPISTAR MRI: multiSlice mapping of cerebral blood flow. Magn Reson Med 1998;40:800–805. 2. Kim SG. Quantification of relative cerebral blood flow change by flow-sensitive alternating inversion recovery (FAIR) technique: application to functional mapping. Magn Reson Med 1995;34:293–301. 3. Kvistad KA, Rydland J, Vainio J, et al. Breast lesions: evaluation with dynamic contrast-enhanced T1-weighted MR imaging and with T2*-weighted first-pass perfusion MR imaging. Radiology 2000;216:545–553. 4. Mills I, ed. Quantities, units, and symbols in physical chemistry. International Union of Pure and Applied Chemistry, Physical Chemistry Division. Oxford, UK: Blackwell Science, 1989. 5. Prüssmann KP, Weiger M, Scheidegger MB, et al. SENSE: sensitivity encoding for fast MRI. Magn Reson Med 1999;42:952–962. 6. Shellock FG. Reference manual for magnetic resonance safety. Salt Lake City, UT: Amirsys, 2003. 7. Sodickson DK, Manning WJ. Simultaneous acquisition of spatial harmonics (SMASH): fast imaging with radiofrequency coil arrays. Magn Reson Med 1997;38:591–603. 8. Wilke NM, Jerosch-Herold M, Zenovich A, et al. Magnetic resonance first-pass myocardial perfusion imaging: clinical validation and future applications. J Magn Reson Imaging 1999;10: 676–685.

5063_Lee_Ch02pp0029-0094 10/20/05 11:02 AM Page 94

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 95

Interventional Computed Tomography Charles T. Burke

Matthew A. Mauro

Percutaneous interventional procedures using radiologic guidance are commonly performed and have become an integral part of the diagnostic and therapeutic armamentarium of radiologists. In concert with the continued growth and advancement of cross-sectional imaging, recent years have seen further development and refinement of interventional techniques and instrumentation, as well as improvements in cytopathologic evaluation. As a result of these advances, imaging-guided procedures such as percutaneous biopsy and abscess drainage have become widely accepted as safe, accurate, and effective alternatives to more invasive surgical procedures. Fluoroscopy, sonography, computed tomography (CT), and magnetic resonance imaging (MRI) all can be used to guide these procedures. The imaging method chosen depends on multiple factors, including lesion size, location, and visibility, the patient’s medical condition, the personal preference, previous experience, and skill of the radiologist, and the relative cost and availability of the guidance methods. (47,98) Fluoroscopy is relatively inexpensive and readily available, and it provides adequate guidance for many lung (403) and skeletal (176) procedures. Large mediastinal and hilar masses (405), as well as lesions identified in conjunction with contrast studies such as cholangiography, pancreatography (147), and angiography (373), also can be accurately sampled with fluoroscopic guidance. Most percutaneous interventional procedures in the abdomen and pelvis, however, are best performed with sonography or CT guidance, which provide superior direct lesion visualization and three-dimensional information (98), though MR guidance may take on a larger role in the future.

3

Paul L. Molina

Sonography is readily accessible and portable, allowing interventional procedures to be performed at the patient’s bedside. It provides real-time multiplanar imaging and continuous monitoring of needle position in a manner similar to fluoroscopy (47). Ultrasound also can be easily combined with fluoroscopy by performing the procedures within an interventional suite, which is particularly helpful when multiple fascial dilatations and drainage catheter placement are required. Ultrasound’s capability of realtime imaging in virtually any plane offers an advantage over CT in guiding nontransaxial needle trajectories, permitting many procedures to be accomplished more quickly than with CT. Ultrasound also is less expensive and uses no ionizing radiation. However, sonographic guidance generally is limited to procedures on relatively superficial structures, particularly in obese patients, because sound attenuation in soft tissues makes deep lesion visualization difficult. Lesions within or deep to bone or air-filled lung or bowel also cannot be adequately visualized because of the nearly complete reflection of sound from bone or air interfaces. Furthermore, sonographic guidance has limited applicability in patients with surgical dressings and draining wounds (98,302). CT provides precise, three-dimensional localization of lesions and surrounding structures, even in patients with external dressings, open wounds, ostomies, or other devices or pathology affecting the skin surface. CT allows imaging of bowel (air) and bone, as well as orally and intravenously administered contrast agents. Oral contrast material aids differentiation of an abscess cavity from normal bowel. Enhancement from intravenous contrast material provides an

95

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 96

96

Chapter 3

assessment of lesion vascularity and delineates nearby vascular structures. The anatomic detail provided by CT enables precise planning of percutaneous access routes for biopsy and/or drainage. The superb spatial resolution also permits exact needle-tip localization within lesions, even those smaller than 1 cm in diameter. Thus, small structures can be biopsied or aspirated with a high degree of accuracy and low risk of complications (269,366). Another advantage of CT as an interventional guidance technique is its versatility. Because a complete 360-degree view of the patient is provided, multiple entry points, approach angles, and potential needle pathways can be selected. Patient positioning can be changed to create safer, more advantageous access routes, and to maximize patient comfort. Additionally, a wide variety of biopsy and drainage devices can be used. Although CT is extremely informative and versatile, CTguided procedures do have several disadvantages. They require expensive scanning equipment and are more costly and time consuming than procedures guided by fluoroscopy or ultrasound. CT also is relatively less available for guidance purposes at most institutions, because of an often demanding, high-volume, diagnostic CT examination schedule. Other drawbacks include limited access to patients because of the CT gantry and the lack of continuous monitoring of needle insertion and guide wire or catheter manipulation. To address these limitations, some manufacturers offer CT scanners with a wider opening between the table and the gantry, which allows placement of a C-arm for combined CT-fluoroscopic guidance (57,127, 336). In addition, several manufacturers offer real-time CT visualization with CT fluoroscopy. In most cases, however, the direction and depth of intervention can be estimated reliably with conventional CT scanners alone, with minimal need for needle or catheter repositioning. Generally, CT is used to guide interventional procedures only when fluoroscopic or sonographic guidance is unsatisfactory. Lesions readily demonstrated by conventional radiographs (e.g., lung or skeletal lesions, gas-filled abscess cavities) are approached with fluoroscopic guidance. Sonographic guidance is employed whenever a lesion (e.g., a predominantly cystic lesion) and a safe access route can be imaged clearly with sonography. Not infrequently, sonography is combined with fluoroscopy and assisted by CT, thus extending the indications for either fluoroscopic or sonographic guidance. For example, after review of the CT scans and selection of an entry site and pathway to a lesion, a fluoroscopic and/or sonographic study is performed to assess the chosen access route. If the access route is appropriate, the needle or catheter is then inserted under fluoroscopic or sonographic direction. Such a complementary approach using multiple methods of guidance often allows complex biopsies and drainages to be performed safely and effectively. CT is used primarily for guidance of procedures involving anatomic areas such as the mediastinum, retroperitoneum,

and pelvis, when lesions are small (
3 cm in diameter) or located next to major vascular structures, when avoidance of bowel loops is important (e.g., in immunologically compromised patients), and for any lesion that is not clearly demonstrated by either ultrasound or fluoroscopy. CT is also used when initial attempts with other forms of imaging guidance have failed. CT fluoroscopy (CTF) has recently emerged as a useful tool in performing biopsies and other CT-guided interventions. Adding a high-speed array processor to a scanner with slip-ring technology enables the generation of very fast reconstructions, creating the impression of “real-time” imaging. Real-time CT imaging combines the benefits of reduced procedure times of fluoroscopy and sonography with the precise needle localization of conventional CT guidance. This allows for safe needle passage while avoiding vital structures, even when only a small window exists, and facilitates placement of the biopsy needles into lesions subject to respiratory motion. Overall success rates are similar to those of conventional CT-guided procedures (83% to 100%) while reducing procedure times by as much as 30% to 50% (42,67,110,354). The development and use of MRI for direction of percutaneous interventional procedures has been described in a number of reports (10,79,139,193,227–230,268,381). However, clinical experience with interventional MRI is still very limited. MRI holds several advantages over CT as a guidance modality: lack of ionizing radiation, multiplanar imaging, and the ability to provide functional information (193, 334). Early animal studies have shown MRI guidance to be feasible for such complex procedures as PTA, intravascular stent placement, vascular embolization, and transjugular portosystemic shunt (TIPS) creation (34,35,179,193,334,410). However, the current state of MRI guidance is limited because it is costly, cumbersome, and time-consuming, and it does not have significant advantages over sonography or CT as a primary interventional guidance technique. It is possible that as future MRI systems continue to offer faster imaging times, improved physical access, and more versatile instrumentation for biopsy, drainage, and tumor ablation, MRI guidance may develop wider application, particularly for lesions that are seen either exclusively or better by MRI than by other imaging techniques (381).

PERCUTANEOUS BIOPSY Percutaneous image-guided needle biopsy has become the most frequently performed interventional radiographic procedure (269). Its most common use is to confirm the diagnosis of a suspected malignancy, whether primary, metastatic, or recurrent. Less commonly, it is used to diagnose and characterize suspected benign lesions. Percutaneous biopsy also is frequently performed to differentiate neoplastic disease from inflammatory disease, postoperative changes, or post-therapy changes (Fig. 3-1). More recently,

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 97

Interventional Computed Tomography

97

B

A

Figure 3-1 CT-guided biopsy. A: CT scan demonstrates a large mass (m) due to metastatic seminoma lateral to the aorta (AO). B: Scan obtained after 6 months of chemotherapy shows that the paraaortic mass had markedly diminished in size (arrowhead). To determine whether this mass represented residual tumor or fibrotic tissue, a percutaneous biopsy under CT guidance was elected. On a scan performed in the prone position, a metallic marker (arrow) was placed on the skin surface to mark the proposed entry site. C: After advancing the biopsy needle, a CT scan confirms that the tip is in the paraaortic mass. Shadowing (arrow) of the needle distally confirms that the needle tip is within the scan plane.

several authors have indicated that measuring tumor markers from biopsy specimens to enhance diagnostic accuracy and perhaps monitor treatment response may further expand the applications of percutaneous biopsy (202,368). For example, it has been shown that tissue levels of carcinoembryonic antigen (CEA) are much higher than serum CEA levels in patients with colonic carcinoma, a known CEA-secreting tumor (202). Therefore, determining tissue CEA levels can be helpful in patients with cystic or necrotic colonic metastases in whom routine biopsy has yielded inadequate cytopathologic information, or in following the response of patients who have undergone alcohol ablation therapy for colonic liver metastases. By safely providing an accurate histologic diagnosis in a majority of cases, percutaneous biopsy often results in more efficient diagnosis and treatment planning. Substantial cost savings may also be obtained, as successful percutaneous

C C

biopsy can obviate more costly and invasive surgical procedures, decrease the duration of hospitalization, and reduce the number of diagnostic examinations performed (30,254).

Needle Selection A wide variety of needles of varying calibers, lengths, tip configurations, and sampling mechanisms are available for percutaneous biopsy (Fig. 3-2). The particular needle selected depends on the location of the lesion to be biopsied, the amount of tissue required for diagnosis, and the individual preference and expertise of both the radiologist and the pathologist. Other primary considerations include patient size and coagulation status, lesion characteristics such as size, vascularity, and proximity to bowel and major vascular structures, and the relative safety of the planned access route. Biopsy needles can be conceptually divided

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 98

98

Chapter 3

obese patients. Such needles are very flexible; they tend to bend and may easily be deflected off course, necessitating multiple needle passes (21,366). However, with moderate experience, a high degree of accuracy can be achieved (142). Although aspiration needles usually allow a diagnosis of malignancy (especially epithelial tumors such as adenocarcinomas and squamous cell carcinomas) to be made, they often do not provide sufficient material to make a specific malignant or benign histologic diagnosis. In cases in which more specific histologic information is required (e.g., lymphoma), the use of a larger, cutting-type needle is often necessary.

Cutting Needles

Figure 3-2 Biopsy needles.

into two general types: (a) aspiration needles for retrieval of cytologic specimens and (b) cutting needles for retrieval of histologic specimens (273).

Aspiration Needles Aspiration needles are simple beveled needles without a sharpened edge or margin to cut tissue (131). They are thin walled, generally have a small (20 to 22) gauge, and are used primarily for cytologic aspiration. Occasionally, they obtain small pieces of tissue that can be processed for histologic examination. The best-known and most commonly used aspiration needle is the Chiba needle (Fig. 3-2). Small-gauge (20- to 22-gauge) aspiration needles are particularly useful when normal organs (especially bowel) or vascular structures must be crossed to obtain a biopsy. Such needles can traverse bowel without substantial risk and have a minimal likelihood of producing significant hemorrhage when vascular lesions are sampled (47,131). Similarly, a small-gauge aspiration needle generally is chosen when a biopsy must be performed in a patient with abnormal coagulation factors. Excellent results have been obtained using CT-guided fine-needle aspiration biopsy, with an overall diagnostic accuracy of 90% reported by several groups of investigators (87,142,184,366). However, the achievement of such a high accuracy rate requires the availability of an experienced cytopathologist. The critical role of the pathologist in obtaining an accurate diagnosis cannot be overemphasized, and cooperation between radiologist and pathologist should be fostered whenever possible (21). The main disadvantage of the thin-walled aspiration needles is the relative difficulty in directing the needle to the lesion, especially with small, deep-seated lesions in

Cutting needles are generally larger needles with a modified sharpened tip or cutting gap on the side (131). They are used primarily for retrieving specimens for histologic analysis. Cutting needles with a modified sharpened tip, socalled modified aspirating needles (e.g., Franseen, Turner, Madayag, Greene), combine the safety of a “skinny” needle with the ability to obtain a tissue core for histologic diagnosis (126). These thin-walled needles have a variety of tip configurations (see Fig. 3-2) and are available in 18-, 20-, and 22-gauge sizes. The Franseen needle has a trephine tip with three cutting teeth that detach pieces of tissue when the needle is rotated (91). This needle is easy to steer and yields excellent cores of tissue (9,66). Along with the Chiba aspiration needle, it is one of the more commonly used needles for CT-guided biopsies. The Turner needle has a sharp cutting cannula with a 45-degree beveled tip, which cuts a core of tissue that remains in the needle when the needle is rotated (209,372). The Madayag and Greene needles both have 90-degree bevels, each with a different type of pencil-point stylet. The stylet of the Madayag is conical, whereas that of the Greene needle is faceted (158). These needles theoretically are easier to direct because of their pencil-point tips, but they may yield less diagnostic tissue than needles with a more acutely angled bevel, such as the 25-degree beveled Chiba needle (9). The Rotex needle also has a uniquely designed needle tip, but unlike the needles described thus far, it does not require aspiration. The needle consists of an inner stylet with a tapered screw tip and an outer cannula with a distal beveled cutting edge (282). The needle winds tissue around the screwlike stylet, which is designed to hold tissue fragments. The outer cannula then slides over the stylet, cutting and trapping the tissue specimen. In one laboratory study evaluating sampling of cadaveric liver tissue by 30 biopsy needles (9), the Rotex needle was found to yield relatively small specimens with notable crush artifact and poor histologic preservation. A more recent in vitro study comparing the Rotex needle with a 30-degree beveled Chiba needle found no statistically significant difference in effectiveness between the two needles, but it noted that the Rotex needle was more expensive and cumbersome to use

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 99

Interventional Computed Tomography

(251). Nevertheless, the Rotex needle has been recommended specifically for lung biopsies, and excellent clinical results with tissue retrieved from both malignant (277) and benign (355) pulmonary lesions have been reported. Some cutting needles use end- or side-cutting mechanisms to actually slice a core of tissue, yielding an excellent histologic specimen. The side-cutting TruCut needle (see Fig. 3-2), available in sizes as large as 14 gauge, is a typical example. It has an outer cutting cannula and an inner slotted needle. After insertion of the needle, the cutting cannula is slid over the needle, slicing the specimen and holding it in the slot of the needle. The cannula also prevents loss of the specimen during withdrawal. Both the Westcott and the Lee needles are similar in design to the TruCut needle but have a smaller gauge. The end-cutting, Menghini-type needle has been used for liver biopsies (136) and for tumors that are very “hard,” especially those within bone (131). A stop on the shaft of the needle can be adjusted to the depth of the lesion. The needle is inserted to the predetermined depth and withdrawn in a single motion (258). Although opinions differ regarding the optimal needle caliber and tip configuration for percutaneous biopsy, it has been shown that more diagnostic tissue can be obtained for histologic analysis with needles of progressively larger bores (9). In an in vitro study in cadaver livers, better tissue retrieval was achieved with 16-gauge needles than with 22-gauge needles (409). However, because of potential bleeding risks, we generally reserve large-gauge (14- to 19-gauge) cutting needles for cases that require histologic information and in which the lesion can be reached without traversing bowel or major vascular structures. If only cytologic information is needed, such as when a biopsy is performed to confirm metastasis or recurrence of known malignancy in a patient with a previously known cell type, then a small-gauge aspiration needle is usually employed for confirmation. Furthermore, if a lesion is known to be hypervascular, biopsy with a cutting needle should be avoided.

Automated Needle Devices Since the mid 1980s, manufacturers have developed a number of disposable and reusable automated biopsy devices, or “biopsy guns,” that can routinely provide cores of tissue for histologic analysis (22,36,50,81,153,233,255, 259,298–300,308). Most of these devices use a spring-activated mechanism to fire a side-notch, TruCut-type needle, although other needle designs (e.g., end-cut, full-cut) are also available. They come in a variety of sizes ranging from 15 to 21 gauge. The Biopty biopsy gun (Bard, Covington, GA) is probably the best known of the automated devices. Initially used primarily for transrectal ultrasound-guided prostate biopsy, it has undergone modifications for use with CT (300) and subsequently has been used for biopsy throughout the body (22,36,50,81,233,298–300,308). It is a hand-held, springloaded, double-throw device that automatically triggers

99

sequential firing of both components of a TruCut-type cutting needle, the inner notched trocar first, followed rapidly by the outer cutting cannula. Many of the other automated biopsy devices use a similar double-throw, or two-stage, biopsy action. There are also several single-throw devices that automatically advance only the outer cutting cannula, which is fired after first exposing the side-notched inner trocar. This is accomplished by either cocking the gun, which withdraws the overlying cannula and reveals the biopsy notch of the trocar, or by manually advancing the inner trocar to expose the side notch (310). Biopsy guns can differ with respect to how far the needle is advanced (biopsy depth or throw length) when the device is activated. Although many are set at a single predetermined biopsy depth, some allow selection of several predetermined depths or even dynamic selection of variable depths (149,170). Additionally, whereas most conventional biopsy guns use TruCut-type needles with a fixed length of the sample side-notch, some use full-cut needles, which are designed to provide a larger volume of tissue. Full-cut needles collect tissue along the entire advanced area of the outer cannula, thus filling the inner volume of the needle. Initial in vitro and in vivo evaluation of a full-cut biopsy gun (Autovac; Angiomed, Karlsruhe, Germany) suggests that specimens obtained with a full-cut-type biopsy gun may be of better quality and quantity than those obtained with a conventional TruCut-type biopsy gun (307). To facilitate safe and proper use of these automated devices, it is important to carefully read the product information to be informed of such variables as type of throw (single or double), biopsy depth, sample notch size, and needle type. Close attention also should be paid to the size and weight of the device. Whether disposable or reusable, those that are too long or too heavy can be cumbersome to use for CT-guided procedures. To address these drawbacks, some reusable devices have a detachable outer cutting-needle cannula, which allows unencumbered initial needle localization, with subsequent attachment of the gun for the biopsy. The cannula can be left in place for additional needle passes, although the act of repeatedly attaching the gun to the needle can be somewhat awkward. Alternatively, some authors advocate the use of a separate coaxial guiding needle through which disposable (259) or reusable (36) automated devices can be repeatedly employed. Use of coaxial guiding needles allows multiple specimens to be obtained with a single localizing pass and provides a route for complementary aspiration biopsy and track embolization (36,259). The automated biopsy devices have been used successfully to perform biopsies throughout the body (22,36,50, 81,170,233,259,298–300,307,308,310,335) and are reported to have advantages over nonautomated needles (22,151, 152,154,259,298). Because they use a simple, standardized action, rather than the multiple, nonuniform motions of conventional aspiration biopsies, these devices can more consistently provide high-quality histologic samples free of significant crush artifact or obscuring blood, even

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 100

100

Chapter 3

from relatively small needles. Their use of cutting needles provides greater yields of diagnostic tissue, allowing for reduction in the number of needle passes required and more prompt, definitive pathologic evaluation. Because of the relative ease of handling and interpretation of specimens obtained by biopsy guns, these devices have been said to lessen the dependence on a readily available, highly experienced cytopathologist (22,259,298). By obtaining a core of tissue, the architectural pattern is retained, allowing for more accurate subtyping of metastatic lesions (364). As a result of reducing the number of needle passes and the need for immediate cytopathologic services, overall procedure time may be shortened compared with manual biopsy techniques (298). The rapid, smooth, clean needle cut afforded by automated devices also reduces patient discomfort (298) and makes it simpler to obtain tissue from organs, such as the pediatric kidney (308). Furthermore, the procedure is made safer by reducing both the number of required needle passes and the amount of time that the needle is in the target organ or lesion (22,233,298,308). Because of the reported advantages over conventional manual techniques, some authors now use automated biopsy devices exclusively for all percutaneous biopsies (298). With continued refinements in design and technical use of these devices, there appears to be potential for safer, more effective image-guided biopsy with further improvements in sensitivity and specificity of both benign and malignant (e.g., lymphoma, sarcoma) diagnoses (22,259,311).

Technique Percutaneous CT-guided biopsy is a safe and highly accurate procedure. In one prospective study of 1,000 CTguided biopsies, a sensitivity of 91.8% and a specificity of 98.9% were achieved, with a positive predictive value of 99.7% (401). The complication rate in this large series was only 1.1%, and the data showed that many different needle types and biopsy techniques can be equally effective in obtaining a diagnosis. Because of the very low complication rate, we routinely perform many CT-guided biopsies on an outpatient basis, unless the patient is hospitalized for another reason. Patient preparation consists of fully discussing the procedure with the patient and obtaining standard informed consent. Patients are placed on a clear-liquid diet on the day of the biopsy to minimize the risk of aspiration if emesis occurs. Premedication with parenteral sedatives or analgesics generally is not required except in young children or in patients with severe anxiety. The only preprocedural laboratory values that are required routinely are a prothrombin time (PT), partial thromboplastin time (PTT), and platelet count. If the platelet count is low (
50,000), platelets can be administered immediately before the procedure. If the PT or PTT is elevated, corrective measures, such as administering fresh frozen plasma, coagulation factors, or vitamin K, should be attempted before proceeding with the biopsy (273).

The first step in CT-guided biopsy is a careful review of the patient’s history, clinical and laboratory findings, and previous diagnostic studies, including a complete CT scan of the area to be biopsied. If contrast enhancement of the diagnostic CT scan is suboptimal, a repeat CT scan with a bolus of intravenous contrast material should be obtained to assess the vascularity of the lesion and surrounding structures (131). The diagnostic CT scan is used to determine the optimal patient position (e.g., supine, prone, decubitus), needle entry site, and pathway to the lesion. Generally, the supine position is easier for patients to maintain than the prone or decubitus positions. The needle pathway is chosen to minimize traversal of vascular structures, bowel, or uninvolved normal tissue. It is also preferable to avoid major nerves or muscles, because traversal of such structures is painful and makes cooperating difficult for patients. The type of information sought from the biopsy (i.e., cytologic, histologic, or both) should also be reviewed with the patient’s referring physician to aid in selection of the type of needle and the sampling technique. The entire procedure can be greatly expedited if such decisions are made prior to placing the patient on the CT scanning table. When prebiopsy planning is complete, the abnormality to be biopsied is relocated on a series of scans obtained at levels preselected from the diagnostic CT study. This can be done relatively quickly by using the longitudinal scan (e.g., Topogram, Scoutview), which is referenced to the transaxial CT slices, and by referring to normal anatomic landmarks, such as the xiphoid or the top of the kidney. It is important for the patient to maintain a consistent phase of respiration during localization and biopsy, particularly for peridiaphragmatic lesions. End-tidal volume (“breath in, breath out, relax, and stop breathing”) seems to be the most reproducible respiratory pattern and is usually well tolerated by the patient. Once the appropriate level for biopsy is found, the needle entry site is selected and its distance from the midline is measured on the CT console. Using the CT laser localizing light, which delineates the selected transaxial level on the surface of the patient, a radiopaque marker (e.g., buckshot, needle) is placed on the skin entry site (see Fig. 3-1). A repeat CT scan is performed to confirm that the marker location is appropriate for needle entry. The skin entry site is then marked with indelible ink and the radiopaque marker is removed. The distance to the lesion and the angle of entry are planned on the CT monitor using the line cursor function to simulate the needle path. Ideally, the shortest straight line from skin to lesion is chosen, although a longer needle path may be selected to avoid critical structures. The skin is cleansed with an appropriate prepping agent, such as povidone-iodine solution (Betadine) and sterilely draped. A local anesthetic agent such as lidocaine (1%) is administered in the skin and subcutaneous tissue, followed by deeper infiltration down to and including the pleura, peritoneum, or capsule of the organ (e.g., liver) while the

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 101

Interventional Computed Tomography

patient suspends respiration. Adequate anesthesia is especially important in very anxious patients, as they are less likely to cooperate if they feel significant pain. If necessary, appropriate doses of Valium or Versed can be given intravenously. During administration of local anesthesia, the underlying tissues are carefully palpated to ascertain the location of potentially obstructing skeletal structures (e.g., ribs, scapula, transverse processes), so that they can be avoided. Occasionally, it is useful to repeat the scan with the anesthesia needle in place to confirm correct entry site and angle. After local anesthesia is obtained, a small cut (3 to 5 mm) is made in the skin entry site with a no. 11 scalpel blade, and the biopsy needle is inserted into the subcutaneous tissue. If necessary, incremental adjustments in needle angulation are made and their adequacy assessed on sequential scans. Once the correct angulation is achieved, the patient is instructed to suspend respiration in the same phase as that used for scanning, and the needle is advanced to the premeasured depth. A variety of depth markers can be used, including a needle stop, sterile tape, or even the protective plastic needle sheath cut to the appropriate length (49). When advancing the needle (especially the thin-walled 20and 22-gauge needles), it is important to maintain a straight course to facilitate accurate placement of the needle into the lesion. A slight deviation in needle course at skin level produces a larger deviation within the body. One hand should be placed at the skin to stabilize and steer the needle, while the other hand is placed on the needle hub and used to advance the needle. A gritty sensation or increased resistance may be felt when a solid lesion is entered. After needle advancement, the patient is instructed to breathe and the needle is allowed to swing free with respiratory motion. If the needle is held or immobilized while the patient moves or breathes, organ movement against the needle creates a lacerating effect. Each time the needle is moved or fixed, the patient should suspend respiration. When the needle is in place, a scan may be obtained to confirm the location of the needle tip. If the needle track is angled, multiple sequential transaxial scans may be required to document the exact location of the needle tip. Alternatively, a longitudinal scan (e.g., Topogram, Scoutview) can be performed to locate the appropriate scan plane for the needle tip. The use of spiral CT may decrease the time required to find the needle tip compared with conventional CT (350). This is because of the ability of spiral CT to image a large volume of tissue rapidly and to eliminate respiratory misregistration, which can be a problem with conventional multislice CT. The key to recognition of the needle tip on CT is the identification of an abrupt, squared end, often associated with a black shadowing artifact (Fig. 3-3) in the adjacent soft tissues (132). Visualization of a distinct feature of the needle, such as a notch near the needle tip, may also be helpful (415). If the needle tip is not within the lesion, it cannot be easily redirected by manipulating the needle hub. It is generally best to remove the needle entirely and reintroduce it.

101

Figure 3-3 Biopsy of nonpalpable rectus muscle mass in an obese woman. The needle enters the mass at a slight angle and therefore is not entirely displayed on a single scan. However, the position of the needle tip is known to be in the lesion by virtue of the shadowing artifact (arrow). Diagnosis: metastatic lung adenocarcinoma.

Once the needle tip is documented to be in the lesion, material is sampled by the appropriate technique, depending on the type of needle (e.g., aspiration, cutting, automated) used. The automated devices, or “biopsy guns,” use an automatic spring-loaded firing device that instantly advances and cuts a core of tissue that is retained within the slotted portion of the biopsy needle. With manual cutting needles such as the TruCut, the outer cutting portion of the needle is manually advanced over the inner slotted stylet to slice a core of tissue and trap it within the slot. When using aspiration or “modified-aspiration” needles, the needle is moved in a circular, reciprocating motion while maintaining continuous suction (5 to 10 mL) with a syringe. The biopsy sample is obtained during suspended respiration and generally takes 5 to 10 seconds to perform. Rotating and advancing the needle dislodges cellular material around the needle tip. Suction is maintained throughout the full excursion of the needle, including needle removal. In a study involving bovine livers, it was shown that nearly a 20% larger sample size could be obtained if suction was sustained, rather than released, during removal of the needle (156). Suction is discontinued only if considerable blood is aspirated. Immediately after needle removal, the patient is allowed to resume normal breathing. Several authors have reported good success in obtaining fine-needle cytologic samples without using any suction (83,331). This so-called nonaspiration technique relies on the physical property of capillary action to draw fluid into the narrow lumen of a fine-gauge needle. Compared with conventional aspiration technique, the nonaspiration technique reportedly is easier to perform and provides a greater concentration of diagnostic cells free of crush artifact or obscuring blood (83,331). However, in an in vitro study using whole organs obtained at autopsy, a lower cellular yield

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 102

102

Chapter 3

was found with the nonaspiration technique compared with the aspiration technique (150). Additionally, in a prospective study comparing the two techniques in abdominal biopsy, no significant differences were noted between them with respect to the number of cell clusters or the amount of blood or crush artifact (182). Overall, the aspiration technique was favored for abdominal biopsy except when extremely vascular lesions were encountered (182). Most authors believe that suction is important and that it should be used to maximize the amount of diagnostic tissue obtained. Several studies have shown that maximum tissue can be obtained if 10 mL of suction is applied in combination with needle excursions greater than 4 mm angled into different portions of the lesion (156,188,189). After removal of the specimen, proper cytopathologic fixation is critical to enhancing the diagnostic yield of the material obtained. Optimally, the cytopathologist or a skilled cytopathology technologist is available to come to the radiology department to process the specimen immediately, assess its adequacy, and provide a preliminary reading, thus providing a service similar to that provided to the surgeon (e.g., for frozen section). This guides the decision as to whether or not additional samples are needed and decreases the rate of false-negative and inadequate biopsies (13,169). Although inadequate specimens can be detected and repeat samples obtained, immediate cytologic assessment may not improve accuracy rates or lower complications in all patients (249). Because this service sometimes is not available, and because it is difficult to predict the adequacy of an aspirate from its gross appearance (87), many radiologists empirically obtain two to six samples, especially when using an aspirating needle (66,87). Diagnoses can generally be made in 80% to 90% of patients after two needle passes, with additional incremental diagnostic yields becoming smaller with each subsequent needle pass (66,87). Cytologic specimens can be smeared on glass slides and immediately fixed in 95% alcohol. These slides then can be rapidly stained with toluidine blue and examined by the cytopathologist. Aspirated material remaining in the needle or syringe is flushed out with fixative into an appropriate container and submitted to the pathology department for centrifugation, filtration, and preparation of a cell-block. Cell-block analysis of the aspirate remaining after the smears are made can increase diagnostic accuracy in some patients (33). Histologic material is placed directly in formalin or in a balanced salt solution and delivered to the pathology laboratory to be processed in the standard fashion. If indicated, specimens also can be sent for microbiologic examination or processed for special studies, such as measurement of tissue carcinoembryonic antigen levels (202). Postbiopsy monitoring is usually continued for 2 to 3 hours and should include an expiration chest radiograph or a postprocedure CT scan for detection of pneumothorax if chest or upper-abdominal lesions have been biopsied.

As postprocedure CT scans often are easier to obtain, they can replace expiration chest radiographs as the method for detecting pneumothoraces after CT-guided interventional procedures (274). If performed, a limited number of postbiopsy CT scans should be obtained through the regions of the chest that are at greatest risk of pneumothorax related to the biopsy.

Tandem-Needle and Coaxial-Needle Techniques To accomplish multiple passes into the same lesion without repeating the time-consuming localization process for each pass, two different methods have been described: the tandem-needle method (86) and the coaxial-needle method (125,135). In the tandem-needle method (Fig. 3-4), a smallgauge needle is advanced into the lesion and, after confirmation of its proper placement, it is left in place to act as a guide for subsequent passes using needles of identical length. Additional needle passes are made immediately adjacent to the guide needle in a similar direction and to a similar depth. Often it is unnecessary to rescan between these additional passes, as the initial guide needle is not withdrawn until the final sample is obtained. With this technique, tissue sampling in several regions of the lesion can be expeditiously performed. With the coaxial-needle technique, also referred to as the double-needle technique, a relatively large-gauge (18- or 19-gauge) needle is used as a guidance cannula for a smaller-gauge (21- or 22-gauge) biopsy needle. The largercaliber guidance cannula, which is inserted first, may be either a short needle used to cross only the superficial musculature, or a longer needle of sufficient length to reach the lesion. Once the direction and position of the guidance cannula is correct, the smaller-gauge needle is passed coaxially through it to the lesion to obtain the biopsy (Fig. 3-5). Typically, multiple passes are made with the smaller needle, using the outer needle for guidance.

Figure 3-4 Tandem-needle technique, pancreatic mass. Second needle (arrowheads) was passed parallel to the guide needle.

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 103

Interventional Computed Tomography

Figure 3-5 CT-guided biopsy of subcarinal mass (M) using coaxial technique. Patient scanned prone. A 22-gauge needle (arrowhead) was inserted coaxially through an 18-gauge needle into the mass. The tip of the 18-gauge needle (arrow) is at the edge of the mass, which is contiguous with bulky paratracheal adenopathy (asterisk). Diagnosis: metastatic small cell carcinoma.

When a long needle inserted to the lesion is used as the guidance cannula (long-cannula coaxial method), precise needle localization is required only once and multiple tissue samples can be obtained with only a single puncture through the pleura, peritoneum, or capsule of the organ. Disadvantages are that a large needle is used to traverse the tissue, and the area sampled generally is smaller than when the tandem-needle technique is used (131). The advantage of using a short needle in the superficial structures external to the body cavity as a guidance cannula (short-cannula coaxial method) is that numerous adjustments in angulation can be made without injury to internal structures (16,131). If desired, changes in angulation of the introducing cannula also can be made between passes to guide the biopsy needle into different portions of the lesion for a wider sampling area, much as with the tandem-needle method.

CT Fluoroscopy When using CT fluoroscopy (CTF) for guidance, additional in-room equipment is necessary. A monitor in the CTscanning suite should be positioned so that it is easily visible to the physician performing the procedure. Many of the current in-room monitors come with a last-image hold feature that may be useful when imaging intermittently rather than using continuous CTF. A joystick attached to the table controls table positioning, and a foot pedal is used to activate the scanner. The technique for performing procedures with CTF is a simple modification of the conventional CT-guidance technique. The initial setup is the same as described for conventional CT-guidance technique. The difference arises when the biopsy needle or, if coaxial technique is being used, the introducer needle is initially being advanced into the soft tissues. At this stage, continuous CTF can be used to localize and direct the needle. Real-time visualization, at

103

this stage, ensures that the needle is started on the proper trajectory. Once the needle has been advanced several centimeters, the operator may opt to only obtain intermittent images to confirm needle location, similar to standard CTguidance technique. This has been referred to as the “quick check” method (354), which helps to reduce the radiation exposure to both patient and personnel. When the projected needle path contains only a small window between vital structures, or when targeting lesions subject to respiratory motion, it may be necessary to obtain images more frequently or use continuous CTF while advancing the needle. Using CTF poses the potential risk of large doses of radiation to both patient and in-room personnel (278,354); it is incumbent on the person performing the procedure to try to limit radiation exposure as much as possible. This may be accomplished by limiting the amount of radiation generated and limiting the personnel exposure to scatter radiation. Minimizing the total CT-fluoroscopic time and reducing the CT scanning parameters will limit the total radiation dose. Often it is not necessary to use the same CT imaging parameters used in diagnostic imaging when performing CT-guided biopsies. Rather, the CT parameters may be set to the lowest possible levels that permit adequate visualization of the lesion and any potentially intervening vital structures in the projected needle path. Daly et al. report that using 120 to 135 kVp and 30 to 80 mA provides adequate visualization in most patients (67). Paulson et al. recommend starting at the lowest possible milliampere setting and increasing the mA when visualization is inadequate (301). In a prospective study, they found 78% of procedures could be successfully performed at a 10 mA setting. CTF-guided procedures are unique from ultrasound and from conventional CT-guided procedures in that all inroom personnel will be exposed to scatter radiation and, therefore, need to wear protective clothing, including lead apron, thyroid shield, and lead-lined glasses. A lead drape placed on the patient just caudal to the imaging plane has been shown to reduce radiation exposure to the physician by as much as 200% (157). In addition, reducing the section thickness of each image from 10 mm to 5 or 2 mm can reduce personnel exposure by 50% to 80% (278). The physician’s hand should never be placed directly into the radiation beam. Instead, when it is necessary to advance the needle under continuous CTF, a needle holder should be used. Specially designed needle holders have been developed for this purpose (157,175). Alternatively, sponge forceps or a similar tool may be used.

Angled Biopsy It is generally preferable to keep the needle track perpendicular to the axis of the patient and thus in a single CT scan plane. This simplifies needle placement as well as CT localization of the needle tip in the lesion. Occasionally, an angled approach, with the entrance site at one level and the target lesion at another level, is necessary to avoid crossing the pleura, lung, vital organs, or major vascular structures.

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 104

104

Chapter 3

and the cost and complexity of these devices varies widely.

Complications Figure 3-6 Prone patient, triangulation method: construction of a right triangle (X,Y,Z), with Z at a point directly above the mass (X) to be biopsied. Y is the prospective needle entry site. The needle path length C and angle a can be calculated according to the Pythagorean theorem.

Determination of the correct depth and angle can be made using a geometric approach (14,104,390). This method applies the fundamentals of geometry to calculate the length and angle of the needle path needed to get to a lesion X from a safe skin site Y (Figs. 3-6 and 3-7). The first step is to determine the straight vertical distance from the skin surface to the lesion. The point on the skin directly overlying the lesion is labeled Z, and the lesion is designated X. The distance from Z to X is measured using the computer software and is labeled B. The second step is to review serial scans and determine a safe skin entry point Y for the needle insertion site in the same sagittal plane as the lesion. The distance from Z to Y is labeled A. A right triangle is then constructed. Using the Pythagorean theorem (A2  B2  C2), the length of the third side C, representing the path of the needle, can be calculated. Finally, the angle of entry a can be calculated using trigonometry. The tangent of angle a is equal to the ratio of the opposite side B over the adjacent side A. The angle of the needle to the horizontal will be equal to angle a. Once the length and angle of the needle path are calculated, the biopsy procedure is performed in the usual manner. It is important that the needle stays in the intended sagittal plane and that a goniometer (or other comparable variable angle measurer) be available to allow the operator to obtain the proper needle angle. A simpler application of this approach (104) is to transfer the full-scale distances A (measured directly from the patient) and B (measured by CT software) onto two edges of a large index card. Connecting these two lines, a triangle can be completed with C representing the length of the desired needle path. The angle of the approach also can be measured directly from the index card by measuring a using a goniometer. Alternatively, the triangle can be cut out, turned over, and held by the side of the patient to directly guide the needle path length and angle (Fig. 3-8). This method circumvents the need to use trigonometric tables and is less likely to lead to mathematical mistakes. A variety of other methods also can be employed to guide the needle at the correct angle. These include gantry tilt methods (363,419), hand-held guidance devices (294,315), light guidance systems (92,94,247,276), and even guidance by elaborate, stereotaxic instruments (58,187, 288,289). Each method has advantages and disadvantages,

Complications from image-guided needle biopsy are infrequent and generally minor, particularly when “skinny” (22-gauge) needles are used. Skinny needles can traverse vascular structures, vital organs, or bowel loops without significant clinical sequelae (47,131,366). However, it is prudent to avoid these structures when possible. Avoidance of bowel, for example, often can be accomplished by changing the position of the patient or by using a blunttipped needle that displaces bowel as it is advanced. Injection of carbon dioxide also can be used to move bowel loops from the intended needle path (133). The risk from needle biopsy increases with increasing needle diameter. In a study of 1,000 CT-guided biopsies, complication rates ranged from 0.3% with 21-gauge needles to 3.0% with 15-gauge needles (401). Risk also is increased when large, cutting-type needles are used (275). Nevertheless, complications from large-bore and cutting needles are minimal when adequate precautions are taken and careful CT-guidance technique is employed (130). The most likely potential risk of percutaneous biopsy procedures is bleeding, which may occur immediately after biopsy or be delayed but is usually clinically insignificant (414). The incidence of significant bleeding (i.e., requiring transfusion) is low, occurring in only 1% to 2.9% of patients following renal biopsy (8,74). Highly vascularized lesions impose an increased risk from percutaneous biopsy, but they can be identified prospectively with good bolus CT technique (130). In such cases, skinny needles should be used for biopsy rather than cutting needles. A relative contraindication to percutaneous biopsy is a bleeding disorder. However, administration of appropriate blood products or medications (e.g., vitamin K) can be used to reverse the clotting disorder before biopsy. Other complications have been reported, and the overall complication rate for fine-needle biopsy is approximately 2% (215,356). Most complications are minor, including vasovagal reactions, transient fever, and pneumothorax. Bacterial contamination, bowel perforation, sepsis, and pancreatitis are all rare events, occurring in less than 1% of biopsies. The mortality rate with needle biopsy is estimated to be 0.1% or less, with most of the deaths occurring after biopsies of the liver or pancreas (357). Needle-track seeding also is an exceedingly rare complication, with an estimated occurrence rate of less than 0.01% (357).

Selected Biopsy Sites Lung and Mediastinum Fluoroscopically guided needle biopsy of lung and mediastinal abnormalities is an easily performed and wellestablished procedure. Because of its simplicity and accuracy,

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 105

Interventional Computed Tomography

105

A

B

C

D Figure 3-7 Adrenal biopsy using the triangulation method (see Fig. 3-6). A: Left adrenal mass (M) in a patient with esophageal carcinoma. Patient scanned prone. A directly perpendicular needle track would necessitate crossing the lung. Arrow, skin site, above mass. B: Therefore, a safe skin entry site (Y) is selected 6 cm lower (A  6 cm). A metallic marker indicates the proposed skin entry site (arrow). C: The distance (B) from the skin (Z) to the mass (X) is calculated using computer software. This allows both the length of the needle path (C) and the angle of the entry to be calculated. D: Scan documenting position of the needle tip in the mass (arrow). Diagnosis: metastatic esophageal carcinoma.

it is the method of choice for biopsy of most chest lesions. However, CT guidance is useful for biopsy of selected lesions, including mediastinal lesions, lesions not easily seen with fluoroscopy (e.g., those that are small, ill defined, or apical) (Fig. 3-9), and lesions adjacent to major vascular structures (2,55,89,95,96,117,119,124,142,349,377,382). Computed tomographic guidance also is useful in patients with suspected lymphoproliferative or benign disease, in whom precise placement of large-bore cutting needles may be necessary to obtain adequate tissue for diagnosis (124). Additionally, CT can be used for preoperative localization of small, nonpalpable, peripheral pulmonary nodules to facilitate their thoracoscopic resection (88,208,306,339,369). Percutaneous CT-guided localization techniques using either spring hook-wires (88, 306, 339, 369) or methylene blue staining of pulmonary nodules (208) have been described.

Reported accuracy rates for CT-guided biopsies of the chest range from 83% to 100% (89,96,119, 349,377). In one large series (377), CT-guided biopsy was diagnostic in 124 (83%) of 150 difficult thoracic cases, including 86 (80%) of 107 pulmonary lesions, 28 (90%) of 31 mediastinal masses, and 10 (83%) of 12 pleural lesions. Diagnosis was established in 107 of 117 proved malignant lesions and in 17 of 26 proved benign lesions (377). Lesion size and location have been shown to influence accuracy rates (129,256,400,418), with lesions less than 1 cm in size having lower reported rates of accuracy (77% to 84%) (287,400,418). One study also reports decreased accuracy for lesions greater than 5 cm in size, presumably due to higher incidence of obtaining a necrotic sample in larger lesions (418). Likewise, lesion location may influence biopsy success and likelihood for complications. Small lesions

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 106

106

Chapter 3

A

B Figure 3-8 Triangulation method, direct guidance. A: The mass (A), skin site above the mass (B), and skin entry site (C) are measured from the CT scan and transferred to an index card. B: The card is then turned over and held by the side of the patient to directly guide the angle and depth of biopsy.

(
2 cm) in a subpleural location, in particular, have been shown to result in lower diagnostic rates as well as higher incidence of complications (129,417). CT-guided biopsies of the chest usually are performed using either a single-needle or coaxial-needle technique (125,382). Because of respiratory motion, it is critical that the patient suspend respiration in the identical phase used for the localizing scan. In general, small-gauge (20- or 22-gauge) aspirating needles are preferred, although large-bore and cutting needles can be used for accessible mediastinal and pleural lesions, particularly when either lymphoma or unusual tumors such as mesothelioma are suspected (124). Any of the automated biopsy devices also can be used when appropriate.

Whenever possible, vessels, bronchi, and bullae should be avoided to minimize complications. For example, when using an anterior parasternal approach, a site must be chosen either immediately lateral to the sternum or greater than 2.5 cm from the sternal border to avoid hemorrhagic complications from injury to the internal mammary vessels (Fig. 3-10) (114,115). Other complications such as pneumothorax may be decreased by minimizing the amount of aerated lung traversed (141). In a review of 131 consecutive CT-guided lung biopsies, the rate of pneumothorax was 46% in patients in

Figure 3-10 CT-guided biopsy of anterior mediastinal mass (M). Figure 3-9 CT-guided biopsy of right apical mass (M) that was ill defined and difficult to see with fluoroscopy. Patient was scanned prone. A 20-gauge needle was inserted medial to the scapula (S) into the mass. Diagnosis: squamous cell carcinoma.

An 18-gauge needle was inserted into the mass via a right anterior parasternal approach immediately lateral to the sternum and medial to the right internal mammary vessels (arrow). Diagnosis: metastatic lung adenocarcinoma. AA, ascending aorta; DA, descending aorta; PA, pulmonary artery; S, superior vena cava.

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 107

Interventional Computed Tomography

whom the needle traversed aerated lung, compared to 0% in those in whom aerated lung was not traversed (141). Similarly, during biopsy of mediastinal lesions, the incidence of pneumothorax may be minimized by using approaches that avoid penetration of the lung or visceral pleura. These include suprasternal or transsternal biopsy of anterior mediastinal lesions and direct mediastinal paravertebral biopsy of middle and posterior mediastinal lesions with the patient in the prone or decubitus positions (Fig. 3-11) (28,64,162). To facilitate maintenance of an extrapleural approach, the extrapleural space can be widened by injection of saline and/or anesthetics (e.g., lidocaine) into the mediastinal tissue, resulting in lateral displacement of the parietal mediastinal pleura (337). Extrapleural injection of saline into subpleural soft tissues also has been used to provide a safe access route for selected peripheral pulmonary lesions (185). Pneumothorax is the principal complication of CTguided chest biopsies, occurring in 25% to 43% of patients (119,349,377), somewhat higher than the average rate of approximately 25% to 30% after fluoroscopically guided thoracic biopsy (403,405). The difference is likely the result of the generally greater difficulty and longer procedure time of cases selected for CT biopsy (95). Factors that have been associated with increased risk of pneumothorax include increased patient age, no previous thoracic surgery, supine positioning during biopsy, and history of smoking (59,90,178). Most patients with pneumothorax require no therapy, but 5% to 18% may require placement of a chest tube (119,349,377). Equipment for immediate chest tube placement should be available whenever a chest biopsy is performed. Once a pneumothorax has occurred, additional biopsies are much more difficult because the position of the lesion often is changed relative to the skin entry site

107

Figure 3-11 CT-guided biopsy of ill-defined mediastinal mass (M) encasing the descending aorta (AO). Mediastinal lesions such as this are difficult to see with fluoroscopy. In addition, major vascular structures (e.g., aorta, heart) are directly adjacent, necessitating accurate needle placement. With the patient in prone position, a 22gauge Chiba needle was inserted into the mass extrapleurally using a left paravertebral approach. The extrapleural space has been widened by injection of saline. Note lateral displacement of parietal mediastinal pleura (arrow). Diagnosis: squamous cell carcinoma.

and because the collapsed lung tends to move away from the advancing needle tip. Hemoptysis also can occur after CT-guided thoracic biopsy (Fig. 3-12), although much less commonly than pneumothorax. The reported incidence of hemoptysis in most series is less than 5% (183). It almost always is self-limited.

B

A Figure 3-12 CT-guided lung biopsy complicated by hemoptysis. Patient has had previous right upper lobectomy for bronchogenic carcinoma. A: A 20-gauge Chiba needle was inserted into a 1-cm right lung nodule. Sutures are present along the posterior aspect of the nodule (arrow) at the site of previous pulmonary resection. Diagnosis: recurrent carcinoma. B: Postbiopsy scan demonstrates parenchymal hemorrhage (H) along the entire length of the needle track. The patient developed self-limited hemoptysis shortly after the biopsy procedure.

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 108

108

Chapter 3

Figure 3-13 CT-guided biopsy of hepatic lesion (arrows) with a 14-gauge TruCut biopsy needle. Previous biopsies using smallergauge aspiration and cutting needles had been nondiagnostic. Diagnosis: focal nodular hyperplasia.

Liver “Blind” percutaneous liver biopsy is sufficient for the diagnosis of most diffuse hepatic diseases, with reported accuracies of 80% to 100% (408). However, for focal hepatic lesions, the diagnostic yield of blind hepatic biopsy is much lower—20% to 70% (56,408). With either ultrasound or CT guidance, focal hepatic lesions can be sampled with an accuracy of 83% to 100% (7,87,136,158,209, 236,283,291,408,409,424). In general, we prefer ultrasound guidance, when possible, because of its ready availability and lower cost. However, CT guidance often is necessary when the lesion is poorly visualized by ultrasound or is in a difficult anatomic location. By using CTF, the procedure time may be reduced so that it is similar to the procedure time for ultrasound guidance (340). The diagnostic accuracy of CT-guided liver biopsy varies according to the cell types sampled and the types of needles used (7,87,136,158,209,236,291,408,409). In most cases, cytologic samples obtained with a 20- or 22-gauge aspiration needle are adequate to confirm metastatic disease or recurrence of a previously diagnosed tumor. If the histologic diagnosis of a lesion is not known, and the lesion is avascular and safely accessible, larger (14- to 18-gauge) cutting needles can and should be used (Fig. 3-13). The diagnostic tissue yield with 18-gauge cutting needles is higher (90% to 95%) than with smaller-gauge aspiration needles (80% to 85%), without a significant increase in complication rate (236,291,424). Histologic samples also can be consistently obtained with any one of a variety of automated biopsy guns or devices. Whenever possible, it is best to choose a needle path that interposes a cuff of normal liver between the capsule and the lesion to be sampled. This cuff of normal liver tissue can help prevent leakage of blood or tumor cells into the peritoneum. If the lesion appears necrotic, one should

attempt to sample its periphery, where more viable tissue is present. For lesions located high within the liver, an angled approach with an entry site below the insertion of the diaphragm may be necessary to avoid the pleural space. The presence of ascites should not be considered a contraindication to liver biopsy. The complication rate after CT- or ultrasound-guided liver biopsies in the presence of ascites has been shown to be no higher than after similar biopsies performed in the absence of ascites (272). Significant complications after CT-guided liver biopsy are rare (approximately 1% to 2%) and consist mainly of hemorrhage (7,136,236,291,408). Sampling of vascular lesions including hemangiomas and hepatomas increases the risk of bleeding (31,136), underscoring the need to assess lesion vascularity on dynamic contrast-enhanced CT scans before biopsy (130,291). If a lesion is vascular, biopsy with a cutting needle should be avoided. To minimize complications in such cases, the lesion should be sampled judiciously with a small-gauge aspiration needle only. Use of a hemostatic protein-polymer sheath, embolization of biopsy tracks with gelatin particles or metallic coils (6,53,62) (Fig. 3-14), and use of transvenous biopsy techniques, also have been advocated to reduce hemorrhagic complications in high-risk patients.

Pancreas The most frequent indication for percutaneous biopsy of the pancreas is to establish a definitive diagnosis of pancreatic carcinoma, at times differentiating it from subacute or chronic pancreatitis. Because most patients with pancreatic carcinoma are unresectable at the time of detection, successful percutaneous biopsy often obviates surgery, except of course in patients who require surgical biliary and/or gastric diversion (e.g., patients with an obstructing mass in the head of the pancreas). In these individuals, open biopsy generally is performed at the time of surgery. In patients managed with percutaneous or endoscopic biliary drainage, CT-guided or fluoroscopically guided biopsy can be performed after the drainage tube has been placed (Fig. 3-15). Many patients referred for pancreatic biopsy have CT findings indicating the presence of metastatic disease (e.g., liver lesions) in addition to the primary mass in the pancreas. In such patients, CT-directed biopsy of potential metastatic lesions in the liver should be performed first, because they generally provide a higher diagnostic yield compared with pancreatic biopsy (304). Furthermore, documentation of hepatic metastases in patients with pancreatic carcinoma has important clinical implications (234). Pancreatic biopsies are most often performed using an anterior approach with the patient supine. At times, placing the patient in a lateral oblique or decubitus position may help to move intervening structures out of the way. Because of the need to cross stomach or bowel with the needle in most cases, small-gauge (20- or 22-gauge) aspiration needles usually are used. However, pancreatic carcinoma is difficult

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 109

Interventional Computed Tomography

109

A

B Figure 3-14 CT-guided biopsy of hepatic lesion with gelatin embolization. A: 17-gauge coaxial introducer needle advanced to the lateral margin of the lesion. The lesion is largely obscured by artifact from the needle (curved arrow). B: After biopsy, the tract was embolized with gelatin pledgets. Air within the gelatin pledgets is seen on post-biopsy can (arrow). Diagnosis: metastatic adenocarcinoma.

to accurately diagnose from cytologic samples alone (252), and the additional use of modified aspiration needles (158,209,409) or multiple passes, or both, may improve diagnostic tissue yield. It is also noteworthy that even with open surgical biopsies of pancreatic carcinoma, false-negative diagnostic rates of up to 14% have been reported (211). The accuracy rates of CT-guided pancreatic biopsies have been reported to range from 60% to 94% (27,71,75,87, 142,158,209,366,409). Generally, these rates are somewhat lower than those reported for CT-guided biopsies at

Figure 3-15 CT-directed biopsy of the body of the pancreas in a jaundiced patient following percutaneous biliary drainage. A 22-gauge Chiba needle was passed directly through the overlying stomach (ST) into the enlarged pancreatic body. Diagnosis: adenocarcinoma. Arrow, biliary drainage catheter; GB, gallbladder containing contrast material; asterisk, incidental left upper pole renal cyst.

other abdominal sites. This is probably because pancreatic carcinomas are often associated with surrounding fibrotic or inflammatory changes, or both, that may obscure tumor location and because small-gauge (20- or 22-gauge) aspiration needles are usually used for pancreatic biopsy (158, 304). Several investigators have demonstrated that largegauge (16- to 19-gauge) cutting needles (27) or biopsy guns fitted with 18-gauge cutting needles (24,81) may yield better results without a significant increase in complication rate. One retrospective review of 211 CT-guided and 58 ultrasound-guided pancreatic biopsies using cutting needles demonstrated a 93% combined CT and ultrasound accuracy in the diagnosis of pancreatic malignancy (27). Accuracy was higher with larger masses (3.0 cm, 92%;
3 cm, 81%) and larger needle sizes (16- to 19-gauge, 92%; 20- to 22-gauge, 85%), and with masses located in the body or tail of the pancreas (93%) rather than the head (84%). Major complications developed in three cases (1.1%). In this study, large needles generally were not used when crossing bowel, particularly when traversing the colon. In 92% of the CT-guided biopsies in this series, intravenous contrast material was administered to better delineate the lesion within the pancreas. If a small pancreatic lesion was better visualized with ultrasound, the biopsy was performed under ultrasound guidance. Such a flexible approach to percutaneous pancreatic biopsy can enhance diagnostic accuracy rates (261). Significant complications resulting from fine-needle pancreatic biopsy are very uncommon (generally
3%), consisting mainly of hemorrhage or pancreatitis (27,71,75, 122,140,265). The incidence of pancreatitis may increase

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 110

110

Chapter 3

when biopsying smaller masses (
3 cm) or when multiple needle passes are made, particularly if normal pancreatic tissue is traversed (265,357).

Adrenal Gland Percutaneous biopsy of the adrenal gland is a valuable alternative to surgery for establishing the histology of an adrenal mass. Most often, CT is used for guidance, although for larger lesions sonography can be used. Major indications for CT-guided adrenal biopsy include assessment of an adrenal mass discovered during staging evaluation of a known malignancy, and characterization of an adrenal mass found incidentally in a patient with no known malignancy (186). Documentation of adrenal metastasis plays an important role in staging and in deciding therapy. Although the adrenal gland is a common site for metastatic disease, an adrenal mass in a patient with cancer has only about a 45% to 55% chance of being a metastasis (20,25,177,402). Just as commonly, such a mass represents a benign process, such as a nonhyperfunctioning adrenal adenoma. Occasionally, CT-directed biopsy also can be used to document metastatic disease in adrenal glands that are normal in size and morphology on CT (292,293). In many cases, noncontrast CT study or MRI using inphase and opposed-phase gradient echo pulse sequences can distinguish between adrenal adenoma and metastasis, obviating percutaneous biopsy (see Chapter 19). In patients without an underlying malignancy, most adrenal masses 3 cm or smaller are benign and usually represent nonhyperfunctioning adrenal adenomas or focal areas of nodular hyperplasia (116,253). In these patients, a more conservative diagnostic course using either MRI or serial CT scanning as an alternative to image-directed percutaneous biopsy usually suffices (20,116,253,352). CT-guided adrenal biopsy has been reported to have an accuracy ranging from 81% to 100% (20,25,144,177,293, 352,402,423). Accuracy usually improves with the use of larger (18- or 19-gauge as opposed to 21- or 22-gauge) needles with no apparent increase in the complication rate (402). In addition, image-guided adrenal biopsy has been shown to have a negative predictive value as high as 92% if normal or benign-appearing adrenal tissue is obtained (352). Adrenal masses can be biopsied with the patient in the prone position using a posterior paraspinal approach that traverses the crura (Fig. 3-16). Although this is usually the easiest and most direct route to an adrenal mass, aerated lung and pleura in the posterior costophrenic sulcus may be traversed by the needle and result in a pneumothorax. To avoid traversing interposed lung, the needle can be angled upward from a more caudal point of entry using various triangulation methods (14,104,390). The angled approach also can be used with the patient in an oblique rather than a prone position. Although angled approaches can avoid intervening structures and reduce the risk of pneumothorax, they are technically difficult and often time-consuming to perform. Sometimes lung can be

Figure 3-16 Direct posterior paraspinal approach to left adrenal mass (M) using CT guidance and coaxial-needle technique. With the patient in prone position, a 22-gauge aspiration needle was placed coaxially through an 18-gauge needle (arrowhead) into the mass. Diagnosis: metastatic lung carcinoma. There is streak artifact over the vertebral body, aorta (AO), and liver from the hub of the 18-gauge needle. K, kidney.

avoided using a direct posterior approach without needle angulation by placing the patient in a lateral decubitus position, with the side to be biopsied dependent (Fig. 3-17) (145). This elevates the dependent hemidiaphragm and reduces respiratory motion on that side. Cephalic needle angulation also can be obviated by placing the patient in a supine or oblique position and using either of two alternative approaches, depending on the side of the lesion. For right adrenal masses, a lateral transhepatic approach through the right lobe of the liver can be used (309). Because the diaphragm is higher laterally than posteriorly, lung generally is not in the path of the needle when this approach is used. For left adrenal masses, a direct approach through the anterior abdominal wall can be used, but traversing the tail of the pancreas is to be avoided, if possible, to reduce the risk of pancreatitis (172). Unfortunately, it is difficult to avoid the pancreas using this approach. In one retrospective review of 48 CT-directed biopsies of left adrenal masses (172), the pancreas was traversed by one or more needles in 32 of 33 cases in which the anterior approach was employed, resulting in acute pancreatitis in two patients (6%). Thus, percutaneous biopsy of a left adrenal mass should be performed from the posterior approach whenever technically feasible. A transsplenic approach to the left adrenal is not recommended, because of the risk of traversing vessels in the splenic hilum. When performing adrenal biopsy, aspiration needles are usually employed because of the frequent need to cross intervening structures between the skin entry site and the adrenal gland. If, however, a safe access route free of vital intervening structures is available, then cutting needles can be used. Complications from percutaneous adrenal biopsy are infrequent and include pneumothorax (292,293,352), hemorrhage (25,293,352,402), and pancreatitis (172).

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 111

Interventional Computed Tomography

111

B

A Figure 3-17 Ipsilateral decubitus position for CT-guided adrenal biopsy. A: CT scan with the patient prone demonstrates bilateral adrenal masses (M). The lower lung is in the projected path for biopsy. K, kidney; arrow, metallic marker on skin. B: The same patient as in (A) in the left lateral decubitus position. The lung has moved up, facilitating a direct posterior approach into the left adrenal mass without the need for needle angulation. Diagnosis: metastatic lung carcinoma.

Major complications occurred in 8 (2.8%) of 277 adrenal biopsies in one series (402). All eight complications were hematomas, one necessitating adrenalectomy. In another series (25), hemorrhage developed in 6 (11%) of 53 patients. However, only two of the patients were symptomatic and required transfusions. An additional potential complication is inadvertent biopsy of a pheochromocytoma, with resultant marked blood-pressure alteration and possible hemorrhage (43,243). If there is a high likelihood of a pheochromocytoma on the basis of clinical and laboratory findings, biopsy should be avoided. However, at least 14% of patients with pheochromocytoma are asymptomatic, and preliminary biochemical studies in these patients may be normal (43). Therefore, during biopsy of adrenal lesions, the possibility of unsuspected pheochromocytoma should be considered, and the radiologist must be familiar with the emergency treatment of hypotensive or hypertensive crises that may occur (43).

98%. In this study, renal mass biopsies were performed with 18- to 21-gauge Chiba needles, and biopsies for medical indications were performed with an 18-gauge cutting needle or a 14-gauge TruCut needle. The rate of serious

Kidney CT guidance for biopsy of focal renal lesions or diffuse renal parenchymal disease seldom is required. In most cases, ultrasound guidance is sufficient and is preferred. CT guidance may be indicated for lesions that are small or are poorly delineated by ultrasonography (406) (Fig. 3-18). The safety and accuracy of CT-guided renal biopsy has been documented by several investigators (275,332,366,406). In one study (366), CT-guided biopsies of 20 renal masses with a 20-gauge needle yielded 100% true-positive results. In another study (275), an accurate diagnosis was obtained in 41 (89%) of 46 patients undergoing percutaneous renal biopsy for renal masses or suspected medical renal disease. When a second biopsy was performed, the accuracy increased to

Figure 3-18 CT-guided renal biopsy in patient with leukemia. 20-gauge Chiba biopsy needle was used to biopsy hypodense renal lesion (arrow). Diagnosis: renal abscess. A, aorta; K, right kidney; S, stomach; GB, gallbladder.

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 112

112

Chapter 3

complications was 6%, consisting of hypotension due to sepsis and retroperitoneal hemorrhage or excessive hematuria requiring transfusion. Other studies have suggested that use of an automated biopsy gun with 18- or 14-gauge cutting needles may be safer (233) and may provide more diagnostic specimens (152) than conventional 14-gauge TruCut needles for medical renal disease. Another potential complication of renal mass biopsy is the occurrence of needle-track seeding. Although there has been a report of needle-track seeding of tumor following biopsy of a renal cell carcinoma with recurrence 20 months later (111), the overall incidence of seeding has been considerably overestimated. In a carefully controlled study of 150 patients with proven renal cell carcinoma (399), there was no significant difference in the 5-year survival rates of a group of 77 patients who underwent diagnostic renal biopsy compared to a control group of 73 patients who did not undergo biopsy. Furthermore, no cases of needle-track seeding were observed. In a more recent study (281), CT-directed renal aspiration biopsy was found to be safe, accurate, and useful in establishing the diagnosis of renal cell carcinoma in 23 patients with disseminated metastases or relative contraindications to nephrectomy.

Retroperitoneum CT is the guidance method of choice for most percutaneous biopsies of the retroperitoneum. Although fluoroscopy or ultrasound may also be used for guidance, CT provides the best visualization for safe, accurate biopsy of retroperitoneal lesions, particularly those in close relationship to the aorta and inferior vena cava. The usual indications for retroperitoneal biopsy include (a) demonstration of metastatic disease in lymph nodes to stage a known neoplasm, (b) diagnosis of lymphoma, and (c) determination of the presence of viable or recurrent tumor after therapy. Retroperitoneal masses generally are approached posteriorly, with the patient lying prone. Occasionally, a direct anterior approach is used (Fig. 3-19). For patients with suspected metastatic disease of a known malignancy, small-gauge aspiration needles usually are sufficient for diagnosis. For unknown or undiagnosed tumors or for lymphomas, larger-gauge cutting needles generally are preferred, because histologic architecture often is important in arriving at a specific diagnosis (21,82,422) (Fig. 3-20). The reported accuracy of CT-guided retroperitoneal biopsy ranges from 65% to 90% (366,401,409,422), depending on the needles used and whether or not lymphomas were included or considered separately in the analysis of biopsy results. Generally, accuracy rates for retroperitoneal lymph node biopsies in patients with metastatic carcinoma are approximately 20% higher than in those with lymphoma (21,401). Many authors believe that large-core biopsy needles are required to subtype lymphoma accurately. However, in a recent review of 102 patients with lymphoma (351), large histologic sections

Figure 3-19 Retroperitoneal lymph node biopsy. Absence of overlying bowel allowed an anterior approach using an 18-gauge Franseen needle. Diagnosis: metastatic cervical carcinoma.

often were not needed to classify the type of lymphoma, and fine-needle biopsies (20-, 21-, or 22-gauge) were just as likely as large-needle biopsies (14- or 19-gauge) to enable both determination of histologic grade and treatment. Therefore, fine-needle biopsy with immediate cytopathologic assessment may be attempted initially in a patient in whom the presence of lymphoma is suspected. If the findings suggest lymphoma, additional fine-needle samples can be obtained for special immunocytochemical studies. If the initial specimens are nondiagnostic, then sampling of tissue with a large needle should be performed. Serious complications are rare after CT-guided retroperitoneal biopsies. Although the major potential complication

Figure 3-20 Retroperitoneal soft tissue biopsy for suspected lymphoma. Using a trans-psoas approach, a 17-gauge coaxial needle was placed into the margin of a large retroperitoneal soft tissue mass. Intravenous contrast was administered to mark the location of the common iliac arteries (CIA). Diagnosis: lymphoma.

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 113

Interventional Computed Tomography

Figure 3-21 Iliopsoas biopsy. Needle is angled along the left iliac bone. Diagnosis: abscess.

is hemorrhage, no significant hemorrhagic episode has been reported (366,401,409,422).

Pelvis The indications for CT-guided pelvic biopsy are similar to those for the retroperitoneum and include the diagnosis of (a) metastatic disease in lymph nodes (290), (b) extranodal metastatic or recurrent disease, particularly in patients who have been treated for colorectal carcinoma (38) or gynecologic malignancy (161), and (c) inflammatory disease (371). CT-directed fine-needle aspiration biopsy has been found to be highly efficient and accurate (96.5%) for assessment of lymph node metastasis in patients with locally confined prostatic carcinoma (290). Likewise, CT-guided biopsy of presacral masses in patients after abdominoperineal resection for colorectal carcinoma is a simple, safe, and valuable procedure for diagnosing recurrence (38). Masses in the upper pelvis usually are approached directly anteriorly with the patient supine, because the bony pelvis prevents a posterior approach (Fig. 3-21). Occasionally, an oblique approach along the iliopsoas muscle is helpful, particularly when using larger-gauge needles. In such cases the needle should be inserted medial to the vessels, nerve, and muscle to minimize discomfort or injury. For presacral or perirectal masses, a posterior approach through the greater sciatic notch with the patient in the prone position is commonly used (Figs. 3-22 and 3-23) (38). Damage to the sciatic nerve usually can be avoided by inserting the needle in the posterior half of the greater sciatic notch (47). No major complications have been reported to date in several large series of CT-directed pelvic biopsies (38, 161,290,371).

Parathyroid Glands CT may be helpful in locating ectopic parathyroid glands in the neck or mediastinum before or after surgical exploration. The CT appearance of parathyroid tissue, however,

113

Figure 3-22 CT-guided biopsy of perirectal nodules (N) after abdominoperineal resection of colon carcinoma. The nodules were not readily palpable on rectal examination. With the patient prone, biopsy was performed using a posterior transgluteal approach through the posterior half of the greater sciatic notch. Diagnosis: recurrent adenocarcinoma.

is not specific. Thyroid tissue, lymph nodes, and thymic remnants can appear similar on CT scans. CT-guided aspiration of suspected ectopic parathyroid tissue, with assay for parathyroid hormone, can improve specificity (78). In one series in which 22-gauge needles were used, a diagnosis of

Figure 3-23 CT-guided biopsy of large pelvic mass. Previous endoscopic biopsy demonstrated villous adenoma. With the patient in a prone position, biopsy was performed through a transgluteal approach using an automated biopsy device. Diagnosis: primary colorectal carcinoma.

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 114

114

Chapter 3

B

A Figure 3-24 Parathyroid aspiration. Patient with persistently elevated serum levels of calcium and parathyroid hormone following neck exploration and subtotal parathyroidectomy. A: CT scan shows right paratracheal mass (arrowheads) just behind the right lobe of the thyroid. B: CT-guided aspiration with assay for parathyroid hormone allowed definitive diagnosis of a parathyroid adenoma. Surgical removal restored normal serum calcium levels.

ectopic parathyroid gland was made correctly in all seven patients without complications (78). Cervical parathyroid masses generally are approached directly anteriorly with a 22-gauge needle (Fig. 3-24). CT provides unequivocal documentation of the needle tip within the lesion. If material cannot be aspirated, gentle irrigation with 1 or 2 mL of saline may be helpful (78). Ectopic mediastinal parathyroid lesions may be similarly aspirated (Fig. 3-25). Although ultrasound guidance probably is more suitable for cervical adenomas, CT guidance is particularly helpful for aspiration of ectopic mediastinal glands.

THERAPEUTIC PROCEDURES Drainage Procedures Computed tomographic guidance for percutaneous aspiration and drainage is a valuable extension of the techniques used for CT-guided biopsy. By far the most common collections aspirated or drained are abscesses. However, CT guidance also can be used to treat a wide variety of other fluid collections, including cysts, lymphoceles, pseudocysts, bilomas, and urinomas.

A

B Figure 3-25 Mediastinal parathyroid aspiration. Patient with hypercalcemia and elevated serum levels of parathyroid hormone. A: Contrast-enhanced CT scan shows a 1.2-cm mediastinal lesion (arrow) between the brachiocephalic artery (bca) and the left brachiocephalic vein (bv). B: CT-guided aspiration with assay for parathyroid hormone allowed definitive diagnosis of a mediastinal parathyroid adenoma.

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 115

Interventional Computed Tomography

115

A

B Figure 3-26 Difficult access route. A: Retrogastric fluid collection (A). The access is difficult because of the surrounding stomach (arrowheads) and spleen (S). B: Catheter placed successfully using CT guidance. Pus was drained.

Abscess Drainage CT is a highly accurate method for detecting abscesses and generally is the imaging technique of choice for their diagnosis. CT can provide detailed three-dimensional display of abscesses and surrounding structures, enabling precise planning of safe, percutaneous access routes for aspiration and/or drainage, as well as immediate evaluation of the degree of success of the drainage procedure after catheter placement. Often, with large, superficial collections, ultrasound can provide similar information. Ultrasound guidance is preferable whenever a collection and a safe access route can be clearly imaged with this technique. Not infrequently, ultrasound is combined with fluoroscopy and assisted by CT, allowing complex drainages to be performed safely and effectively. However, ultrasound frequently is limited by overlying bowel gas, bone, or surgical wounds and drains (98,302). Overall, CT remains the primary planning and guidance technique for percutaneous abscess drainage, particularly for small deep collections and those surrounded by bowel loops or other vital structures.

INDICATIONS AND PATIENT SELECTION The indications for percutaneous abscess drainage (PAD) have expanded considerably in recent years, in large part because of advances in cross-sectional imaging, interventional techniques, and instrumentation. Initially, only relatively stable patients with localized, unilocular, readily accessible, uncomplicated collections were considered appropriate for PAD (3,54,107–109,137). Currently, PAD also is performed for multiple or multilocular abscesses, for abscesses complicated by communications, fistulas, or difficult access (i.e., overlying structures), and for abscesses in critically ill patients (105,180,264,270,378,380,383,389). These expanded applications of PAD can be particularly helpful in patients who are at high risk for surgery (389). Frequently in these

cases, the patient’s condition can be improved so that, if necessary, definitive surgical drainage eventually can be performed. We consider the only absolute contraindication to PAD to be lack of a safe access route. With CT guidance, this is rarely a problem (Fig. 3-26).

Technique Access Route Each patient should have a complete diagnostic CT examination performed before PAD. In general, the shortest, most direct route to the collection is preferred. However, the intended path should avoid bowel loops, major vessels, and the pleural space. Achieving such an access route may require using an angled approach. Diagnostic Aspiration Although CT is a very sensitive method for detecting abscesses, its specificity is relatively low. A variety of other collections can have similar CT appearances, including hematomas, necrotic tumors, loculated ascites, urinomas, bilomas, and lymphoceles. Therefore, before drainage of a suspected abscess, a diagnostic aspiration should be performed. Generally, diagnostic aspiration is done with a 20- or 22-gauge aspiration needle, particularly when the collection is in a difficult location requiring needle passage through normal structures. The technique of needle placement is identical to that described earlier for percutaneous biopsy. Passage through bowel, particularly colon, should be avoided because of the risk of infecting a potentially sterile collection. Aspiration of viscous material often is difficult through a fine 22-gauge needle. If no fluid is retrieved and if the needle is positioned correctly, a largergauge (18-gauge) needle can be inserted for aspiration. If the material from the diagnostic aspiration is obviously purulent, placement of the drainage catheter can begin immediately. However, if the collection is not clearly

5063_Lee_Ch03pp0095-0144 10/13/05 12:03 PM Page 116

116

Chapter 3

infected, a Gram stain can be quickly performed. Gram stain of an abscess specimen typically demonstrates white blood cells and bacteria. Patients receiving antibiotics may have white blood cells but no bacteria, the so-called sterile abscess. Conversely, a Gram stain revealing bacteria but no white blood cells indicates either bowel contents or a severely immunocompromised patient who is unable to mount a leukocyte response (378). Catheter Selection Adequate drainage requires a catheter with a large enough lumen to drain the material freely and multiple side holes to minimize occlusion. Two systems are generally used: pigtail catheter or sump catheter (107,109,135,239,270,380,383,384) (Fig. 3-27). Pigtail catheters are single-lumen catheters with curled distal tips that help provide stability within abscesses and that are relatively atraumatic to abscess walls. Their side holes are located on the inner curve of the pigtail portion of the catheter to prevent occlusion. However, when strong suction is applied, the cavity wall tends to collapse around the catheter, occluding the side holes. Therefore, pigtail catheters usually are reserved for dependent drainage or low bulb suction. Sump (double-lumen) catheters are constructed with two lumens: a small venting air lumen and a larger draining lumen. The double-lumen design allows for concomitant irrigation and drainage. Continuous suction (usually low, intermittent wall suction) can be used to evacuate fluid and debris through the drainage lumen, while room air or irrigant fluid is drawn into the catheter through the sump lumen. This helps to maintain side-hole patency and minimizes the reaccumulation of abscess fluid and debris within the cavity. Sump catheter systems are particularly useful for drainage of thick, viscous collections (181,235,384).

Pigtail and sump catheters come in a wide variety of luminal and side-hole sizes. Some of them also have selfretention devices such as locking loop sutures to help prevent catheter dislodgement. The choice of catheter size and design usually is based on the size, location, and composition of the abscess, as well as on individual experience and preference. In general, the more viscous the collection, the larger the catheter caliber and side holes should be. A 12 to 14 French catheter usually provides adequate drainage for most abscesses (378). For extremely viscous collections such as advanced empyemas, thick hematomas, or necrotic pancreatic tissue, a 16 to 30 French catheter may be necessary. Smaller 7 to 9 French catheters are generally adequate for less viscous collections, such as cysts, seromas, and noninfected pseudocysts (378). Catheter Placement Placement of both types of drainage catheters can be performed by the Seldinger or trocar technique. The Seldinger technique uses standard angiographic methods (Fig. 3-28), whereby a plain or sheathed needle is placed in the collection, usually along the same track used for diagnostic aspiration. After material is freely aspirated, a guide wire is placed into the collection through the needle or sheath. A scan is obtained to confirm that the guide wire is coiled in the collection. The needle is then removed over the guide wire and the needle track is dilated. Finally, the appropriate-sized catheter is threaded over the guide wire and into the collection. The guide wire is removed, and proper position of the catheter is documented with repeat scanning. If desired, once the guide wire is in place, the remaining steps may be performed under CT-fluoroscopic or conventional fluoroscopic guidance. Hybrid CTF systems are available, allowing these steps to be performed without moving the patient. Alternatively, the needle/sheath is removed and the patient is transferred from the CT scanning table to the fluoroscopy room with the guide wire taped in place. If the collection is large and superficial (i.e., simple access route without intervening structures), the drainage catheter may be placed directly by using the trocar technique (Fig. 3-29) (326). With this technique, the catheter is mounted on a sharp metal introducer, and the entire apparatus is placed through the front wall of the collection in one step, similar to placement of a chest tube. The catheter is advanced over the metal stylet into the cavity, and the stylet is removed. A modification of standard trocar puncture also may be used in which the trocar catheter is advanced over an inserted guide wire, rather than the metal stylet (387). The guide wire modification of trocar puncture provides an extra measure of safety and facilitates optimal catheter positioning.

Figure 3-27 Drainage catheters. Pigtail (A), trocar (B), and sump

Management

(C) catheters are shown diagrammatically. The cross-sectional view to the right of each catheter demonstrates the single lumen for the pigtail and trocar catheters and the double lumen of the sump catheter. The air-vent lumen of the sump catheter is indicated (arrow).

After the catheter is in place, the collection is aspirated as completely as possible. Gentle saline irrigation is then performed until the drainage is clear, to facilitate cleansing of

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 117

Interventional Computed Tomography

A

117

B

Figure 3-28 Percutaneous abscess drainage using the Seldinger technique. A:

C

A

C

Diagrammatic representation of a liver abscess (A). The procedure begins with placement of a needle into the collection for diagnostic aspiration. The arrow indicates a strip of sterile tape used as a depth guide. L, liver; ST, stomach; Ao, aorta; K, left kidney; P, pancreas with calcifications. B: A guide wire is then passed through the needle into the collection. The needle is then removed, leaving the guide wire in place. C: A catheter is then threaded over the guide wire, and the guide wire is removed. The catheter is secured in place with suture material. (Adapted from Gerzof SG, et al. Percutaneous abscess drainage. Semin Roentgenol 1981;16:62–71.)

B

Figure 3-29 Percutaneous abscess drainage using trocar technique. A: Aspiration of this diagrammatic retrogastric abscess (A) with a thin needle was done as an initial confirmatory procedure. The needle stop is indicated (arrow). ST, stomach; L, liver; P, pancreas; SP, spleen. B: The trocar unit was advanced to a previously determined distance, with a needle stop in place to prevent puncture of the back wall of the abscess. C: Keeping the stylet in place, the catheter was advanced over the stylet until the drainage tube was well seated in the cavity. The trocar was removed, and dependent drainage was begun. Adapted from Gerzof SG, et al. Percutaneous abscess drainage. Semin Roentgenol 1981;16:62–71.)

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 118

118

Chapter 3

the cavity. The saline should be administered in small volumes (smaller than the amount initially aspirated) and at low pressures to avoid sepsis. A full course of appropriate systemic antibiotics also should be given. Catheters are left to gravity drainage or attached to lowbulb or intermittent wall suction in the patient’s room. Frequent saline irrigation in small volumes may be helpful to keep the catheter patent and to reduce the viscosity of the retained material. Urokinase also has been shown to reduce the viscosity of purulent material and to increase drainage flow rates for all sizes of catheters (296). In select cases (e.g., viscous, multiloculated abscesses; infected hematomas), intracavitary urokinase or tissue plasminogen activator (tPA) administration may be a useful adjunct to PAD and can be given safely during percutaneous drainage of abdominal and pelvic abscesses without significant side effects or alterations in serum levels of coagulation factors (194). Although a rapid clinical response (i.e., defervescence, normalization of white blood cell count) often occurs with the initial drainage, the catheter generally is left in place as long as it continues to drain. Most abscesses require 3 to 10 days of drainage (3,270,383) unless communication to the bowel or ductal (biliary, pancreatic, urinary) system exists. When drainage ceases, the catheter is withdrawn gradually—several centimeters per day—much like a surgical drain. A final CT scan is helpful to confirm complete resolution of the collection before catheter removal (Fig. 3-30). If the patient remains febrile after the procedure, incomplete drainage should be suspected. Repeat CT should

be performed and additional catheters placed if necessary. If drainage tapers and then dramatically increases, a fistulous communication (usually enteric or biliary) may be present (Fig. 3-31). This can be confirmed with a contrast sinogram. If a fistulous track is present, percutaneous drainage can still succeed (180,263,270,279,295,383). However, this often requires extended percutaneous drainage with catheter repositioning or insertion of additional catheters into the vicinity of the fistula as well as dependently within the abscess. For abscesses communicating with bowel, complete bowel rest with appropriate parenteral hyperalimentation accelerates healing.

Results and Complications Proper performance of PAD minimizes manipulation of the abscess cavity, which reduces the risk of procedureassociated sepsis. Advantages of PAD include (a) avoiding the risks of surgery and general anesthesia, (b) saving considerable time and expense, and (c) making nursing care easier because of the closed drainage system. PAD compares favorably with operative treatment in terms of overall clinical success, duration of drainage, and recurrence rate. Success rates of 80% to 90% have been widely reported with PAD (32,54,105,107–109,137,160, 168,235,267,270,380,383). Generally, PAD is most successful if the collection is well defined, unilocular, localized, and free flowing. Such “simple” collections constitute the majority of intra-abdominal abscesses (108). The cure rate for more complex collections, such as pancreatic ab-

A, B

C Figure 3-30 Successful percutaneous abscess drainage. A: Febrile patient 1 month after splenectomy. CT scan demonstrates a large splenic bed fluid collection (arrowheads). B: CT scan following placement of a drainage catheter (arrow) and aspiration of 500 mL of gross pus. The collection has diminished in size. C: Two weeks later, drainage had ceased, and repeat outpatient CT showed no residual collection. The catheter (arrow) was then withdrawn slowly over the next several days.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 119

Interventional Computed Tomography

119

A

B

Figure 3-31 Abscess with fistulous communication. A: CT scan shows a collection in the superior pelvis (arrow) adjacent to the sigmoid colon (curved arrow). The collection contains a large amount of gas, suggesting possible communication with bowel B: Collection percutaneously drained from an anterior approach. C: Contrast injection of the drainage catheter demonstrates fistulous communication with the sigmoid colon (C). A diverticular abscess was the cause.

C

scesses and abscesses caused by enteric or biliary leaks, is lower—32% to 70% for pancreatic abscesses (93,174,205, 360,391) and up to 80% for abdominal abscesses associated with fistulas (180,295). These abscesses are frequently poorly defined, multilocular, and extremely viscous, making complete drainage difficult even with the use of large or multiple catheters. However, a beneficial, temporizing effect can still be achieved by draining most of the infected fluid and debris, thus allowing the patient to undergo subsequent surgery in a more stable condition and with a relatively cleaner operative bed (174,205,360, 389,391). This temporizing, adjunctive role of PAD also has been used

successfully in patients with abscesses secondary to appendicitis (163,164,392), Crohn disease (195,328), and diverticulitis (266,280,329). The overall favorable results for PAD may be related to the high diagnostic accuracy of CT, which facilitates complete delineation of the extent of the collection. In addition, meticulous follow up with repeat CT scans, along with aggressive catheter repositioning and additional catheter placement, is critical to achieving successful drainage. Failure of PAD is usually the result of failure to recognize and respond to loculations, septations, and fistulous communications; premature catheter withdrawal; or an attempt to

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 120

120

Chapter 3

drain a relatively solid collection (e.g., pancreatic inflammatory mass, necrotic tumor) (198,383). Complication rates for PAD generally range from 5% to 15%, with serious complications occurring in less than 5% of cases (32,93,105,108,168,196,198,267,380,383). Minor complications include limited bleeding, skin infection, transient bacteremia, and pneumothorax. Major complications include marked hemorrhage, sepsis, empyema, and bowel perforation. These latter complications occur in less than 5% of cases and result in death in less than 1%. Recurrence rates after successful PAD are approximately 5% to 10% in many reported series (3,168,380,383). These figures compare favorably with those for surgical management of abscesses, which is associated with mortality of approximately 10% to 20% and a recurrence rate of 15% to 25% (3,32,168). In addition, the overall length of hospitalization is shorter with PAD than with open surgical drainage (32,168).

Specific Sites Lung and Mediastinum Most patients with lung abscesses respond to conservative treatment with antibiotics and postural drainage. Nonresponsive patients, however, usually require tube drainage of the abscess or resection of the affected portion of lung. Percutaneous drainage of lung abscesses has been successfully performed in selected patients (i.e., those refractory to antibiotics, or at high surgical risk) using radiologic guidance (11,15,224,297,379) (Fig. 3-32). CT guidance is particularly helpful in choosing the most advantageous route to the abscess. Whenever possible, the drainage catheter should traverse contiguous abnormal lung and pleura rather than normal pulmonary parenchyma, in order to minimize development of an infected bronchopleural fistula or pyopneumothorax (379). It is also best to gain access to the collection with the affected side in the dependent position. This avoids aspiration of pus into the normal lung, which may lead to overwhelming sepsis. Percutaneous aspiration and drainage of mediastinal abscesses also has been successfully performed, mainly in high operative risk patients, or in patients with surgical esophageal anastomotic leaks (118,245). Because of the proximity of mediastinal collections to vital cardiovascular, gastrointestinal, and tracheobronchial structures, CT may be critical in selection of a safe extrapleural access route for aspiration or drainage. Anterior mediastinal collections usually are drained from a parasternal approach, taking care to avoid the internal mammary vessels (114,115). Posterior collections generally are approached paravertebrally. Transesophageal drainage of mediastinal abscesses also has been described in patients in whom a safe, percutaneous access route was not available (245). In one report, eight periesophageal abscesses caused by esophageal perforations were drained through a transesophageal route with minimal patient morbidity (245).

Subphrenic Space The left and right subphrenic spaces are common sites for abscesses after abdominal surgery. When performing drainage procedures of the subphrenic spaces, it is important to avoid traversing the pleural space to prevent potential empyema production. The target for entrance of the needle should be the lowest portion of the abscess, with the needle angled from caudad to cephalad as described earlier for percutaneous biopsy. This permits positioning of the catheter throughout the length of the abscess and helps ensure avoidance of the pleural space. One should also look for extension of subphrenic abscesses into contiguous anatomic areas, such as the subhepatic and pericolonic spaces. If extension is present, these areas may require additional catheter drainage. Success rates of 85% have been reported for percutaneous drainage of subphrenic abscesses (160,267). Complications include empyema and sepsis, both of which are uncommon (
5%). If the pleural space cannot be avoided and an empyema develops, it can be treated rapidly with catheter drainage. Liver Abscesses in the liver usually are approached with an entrance point on the anterior or lateral surface (generally no further posterior than the midaxillary line), taking care to avoid traversing the pleura. As with percutaneous needle biopsy, an angled approach may be necessary, particularly for collections in the dome of the liver. Whenever possible, it is best to choose a path that interposes a portion of normal liver tissue between the capsule and the abscess to be drained. Doing so can help prevent leakage of material into the peritoneum. The success rate for percutaneous drainage of pyogenic liver abscesses ranges from 65% to 90% (12,76,77,106,167). Pyogenic liver abscesses with an intrahepatic biliary communication require drainage for a longer time than do abscesses without communication, but the overall cure rates for the two groups are not significantly different from one another (76). Clinically significant complications are uncommon (approximately 5%) and include sepsis, peritoneal contamination, and production of an empyema. In selected cases, percutaneous catheter drainage also has been used successfully for treatment of amebic liver abscesses (376,385). Specific indications for drainage of amebic liver abscesses include poor response to medical therapy, worsening pain, imminent rupture, and left-lobe abscesses, which have a greater frequency of rupture and complications (385). Intervention also may be necessary to establish the diagnosis of amebiasis in patients with undetermined or false-negative serologic results in whom amebic abscess must be differentiated from pyogenic abscess. Pancreas Pancreatic abscesses are collections of pus located near the gland that contain little or no necrosis. Some authors have reported moderate success (65% to 70%) in draining

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 121

Interventional Computed Tomography

121

A

C

these often viscous collections percutaneously (93,391). When percutaneous drainage of a pancreatic abscess is performed, large-bore catheters often are necessary to drain the viscous material adequately. Close monitoring with follow-up CT examinations and contrast sinograms frequently are necessary, as are catheter exchanges and manipulations. Pancreatic abscesses must be differentiated from infected necrosis, which generally is not amenable to percutaneous therapy. Surgical débridement often is required to remove pieces of necrotic debris that cannot be drained through percutaneous catheters (174,205,360,391). A combined radiologic and surgical approach may be beneficial in selected patients with combined areas of abscess and infected necrosis. It has been recommended that collections in central areas such as the pancreas and lesser sac initially undergo surgical débridement, followed by percutaneous drainage if débridement is incomplete (205). The rationale is that these central collections represent infected pancreatic necrosis, which requires necrosectomy. Collections in more peripheral areas (e.g., pararenal space, paracolic

B

Figure 3-32 CT-guided drainage of lung abscess due to Haemophilus influenzae pneumonia. The abscess was refractory to antibiotic therapy. A: CT scan with the patient supine demonstrates a rounded, low-attenuation abscess (arrowheads) in the left upper lobe. There is adjacent parenchymal consolidation. B: After diagnostic aspiration with the patient in a prone oblique position, a guide wire was advanced and coiled in the abscess (A). Note that a posterior approach through the adjacent consolidated lung parenchyma has been used and that the affected lung is in a dependent position relative to the uninvolved, contralateral lung. C: Patient was scanned prone. A drainage catheter has been advanced over the guide wire into the abscess (A). Purulent material was aspirated, and air now fills the abscess cavity. The patient recovered uneventfully and was discharged home after 1 week of catheter drainage.

gutters) are more likely to represent true abscesses, which are more amenable to percutaneous drainage. Kidney Renal abscesses can be successfully drained percutaneously using radiologic guidance (61,73,197,327). They are best approached posterolaterally through a path that avoids the liver, spleen, and colon. An angled approach with the needle entry site below the level of the 12th rib may be necessary to avoid intervening lung, liver, or spleen when entering upper-pole renal abscesses (61,155). Separate drains may be required for multifocal or multiloculated collections. Perirenal and Pararenal Spaces Abscesses within the perirenal and pararenal spaces are frequent and usually have a lengthy craniocaudal extent as a result of the anatomy of these retroperitoneal spaces. For drainage to be successful, a catheter or catheters should be positioned throughout the entire length of the cavity and placed in a dependent location. Success rates of 61% to 93% have been reported for percutaneous drainage of renal,

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 122

122

Chapter 3

perirenal, and pararenal abscesses (73,197,327). Significant complications are unusual and include hemorrhage, sepsis, and pyopneumothorax. Pelvis Abscesses in the upper pelvis usually are approached directly anteriorly with the patient supine, because the skeletal structures of the pelvis prevent a posterior approach. CT-guided anterior approaches have been used successfully for percutaneous drainage of tubo-ovarian abscesses in patients who are refractory to antibiotic therapy (44,374). When a safe anterior route to a pelvic collection is not available (i.e., in the presence of overlying bladder, bowel, or vessels), a posterior transgluteal approach through the greater sciatic foramen with the patient in the prone position can be employed, with a success rate of 81% reported in one series (37) (Fig. 3-33). When using the transgluteal approach, it is necessary to avoid the sciatic nerve and plexus and the superior gluteal vessels. These important structures course anterior to the pyriformis muscle in the center of the greater sciatic foramen, and they must be identified on preliminary CT scans. The most common complication of this approach is pain, which usually resolves within 24 hours of catheter insertion. However, in approximately 20% of cases, persistent local pain is a significant complicating factor, probably resulting from neural plexus irritation. Because of this, several authors have advocated pelvic PAD by the transrectal or transvaginal route when possible (18,100,191,240). Compared with the transgluteal approach, the transrectal or transvaginal approaches offer increased patient comfort without risk of injury to the sciatic nerve or gluteal vessels.

Pancreatic Pseudocysts The role of imaging-guided percutaneous treatment of pancreatic pseudocysts continues to evolve (93,174,205,391). Generally, the primary indication for percutaneous intervention is infection. If there is a question of an infected pseudocyst, percutaneous aspiration with CT guidance can be accomplished easily and safely with a high degree of diagnostic accuracy (365,391). Like any other abscess, an infected pancreatic pseudocyst should be drained promptly. Surgical drainage of an infected pseudocyst may be difficult, particularly if the pseudocyst wall has not fully matured, a process that generally takes 4 to 6 weeks. In patients with mature (i.e., at least 4 to 6 weeks old) or enlarging pseudocysts and severe abdominal pain, CTguided drainage can be offered as an alternative to surgical drainage (65,159,370,391,393). In one review, 91 (90%) of 101 pseudocysts (51 infected, 50 noninfected) in 77 patients were reported to be cured by means of catheter drainage (393). The mean duration of catheter drainage in this series was 19.6 days, and only six patients required operation after percutaneous treatment. Complications

occurred in 10 (13%) of the 77 patients, four of which were major (bacterial superinfection of previously noninfected pseudocysts) and six minor. External percutaneous drainage of pancreatic pseudocysts is performed in a manner identical to that described earlier for PAD (Fig. 3-34). Several authors also have reported good results using a transgastric approach to pancreatic pseudocyst drainage (23,148,192,286,328,393). With this method, a percutaneous route is chosen that includes both walls of the stomach and the pancreatic pseudocyst. Transgastric pseudocyst drainage mimics surgical cystogastrostomy and allows a mature tract to form from the cyst to the stomach, thus eliminating the risk of a pancreaticocutaneous fistula, which sometimes may occur with routine external percutaneous drainage. If communication of the cyst with the pancreatic duct persists, the cyst contents may still empty into the stomach via the fistulous tract after the drainage tube has been withdrawn. The transgastric approach also allows an internal double-J stent to be placed between the stomach and the pseudocyst without an external drainage tube. Leakage of gastric contents into the peritoneum when using the transgastric approach has not been a frequent complication.

Miscellaneous Drainage Procedures Theoretically, any fluid collection in the body can be drained percutaneously with CT guidance. Examples include pleural effusions and empyema (201,260,353,359, 375,386,404), lymphoceles (407), hydatid cysts (29,262), bilomas (263,396), urinomas, and hematomas. The technique is identical to that described earlier for PAD. Percutaneous Nephrostomy Most percutaneous nephrostomies can be performed easily using either fluoroscopic or sonographic guidance. Although the indications for CT guidance are limited, CT may provide a higher margin of safety in renal transplant patients, very high-risk patients, or patients with a single kidney (138). CT guidance may be helpful in immunosuppressed patients to avoid both the peritoneum and inadvertent bowel puncture (138). The procedure can be done in conjunction with fluoroscopy for nephrostomy tube placement once the guide wire is properly inserted. Cecostomy CT guidance also has been used successfully to place percutaneous cecostomy tubes for cecal decompression in Ogilvie syndrome (idiopathic pseudo-obstruction of the colon) (45,60). This can be accomplished from either an anterior or a posterior approach. A posterior retroperitoneal route may be preferred to avoid crossing the peritoneum. The advantage of percutaneous cecal decompression is that surgery may be avoided in high-risk patients.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 123

Interventional Computed Tomography

A

123

B

Figure 3-33 Transgluteal CT-guided drainage of pelvic fluid collection. A: Initial prone localizing scan shows a fluid collection (fc) located adjacent to the uterus (U) and between the rectum and bladder (B). A 25-gauge needle (arrow) marks the skin entry site. B: A 21-gauge needle was inserted transgluteally through the posterior half of the greater sciatic foramen. Note that the needle avoids the rectum and pelvic blood vessels. The tip of the needle is located at the margin of the fluid collection. C: Scan demonstrating a 0.018-inch guidewire (arrows) coiled within the collection. This confirms satisfactory guide wire location prior to catheter placement.

A

C

B Figure 3-34 CT-guided pancreatic pseudocyst drainage. A: Contrast-enhanced CT scan shows a lobulated pseudocyst (arrows) involving the head of the pancreas. B: The pseudocyst has been completely drained after CT-guided placement of a percutaneous drainage catheter.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 124

124

Chapter 3

Necrotic Tumors A less common use of percutaneous catheterization is drainage of necrotic or infected tumors (1,271). Unlike drainage of standard abscesses, however, percutaneous drainage of tumor is not intended to be a curative procedure. Rather, it is used as a method of providing palliation in patients who are either unable to tolerate surgery or in whom other treatment modalities have been exhausted.

Neurolysis Celiac ganglion block is performed with neurolytic agents such as alcohol or phenol for relief of intractable abdominal pain that is refractory to narcotic therapy. It is most often used in patients with malignant abdominal disease (e.g., pancreatic, gastric, and metastatic cancers), although patients with severe pain related to benign abdominal disease (e.g., chronic pancreatitis) also may be treated (204). CT guidance of the procedure facilitates direct administration of alcohol into the region of the celiac ganglia. The celiac ganglia are the largest of three pairs of sympathetic ganglia that make up the celiac plexus, which is located along the anterior surface of the upper abdominal aorta. Percutaneous neurolysis of the celiac plexus has been performed successfully using either anterior (112,204,317), posterior transcrural (39,134,204), or posterior transaortic approaches (210). The technique for CT-guided percutaneous neurolysis is similar to that for percutaneous biopsy. Preliminary contiguous thin-section scans (5-mm slice thickness at 5-mm intervals) are obtained through the upper abdomen, and the celiac and superior mesenteric arteries are identified. Using either an anterior or a posterior approach, one or two 20-

gauge needles are placed just anterior to the diaphragmatic crus at or between the levels of the celiac and superior mesenteric arteries (Figs. 3-35 and 3-36). After CT documentation that the needle tip is in the appropriate location, extension tubing is attached, and air or dilute contrast medium (approximately 5 mL) is injected through the needle to assess spread around the aorta. The injected air or contrast medium should flow freely and diffuse into the retroperitoneal space around the celiac and superior mesenteric arteries (see Figs. 3-35 and 3-36). A test dose of local anesthetic agent (e.g., lidocaine) also may be injected at this time to assess the patient’s response. If the patient has subjective pain relief, and if adequate spread is observed, 10 to 20 mL of absolute alcohol is then injected through the needle. Before needle removal, 5 mL of saline is flushed through the needle to prevent alcohol from tracking back along the needle path. Pain relief has been reported in 60% to 90% of cases, but this is difficult to grade, because one must rely on the subjective impression of the patient (112,210,317). In one study, CT-guided celiac ganglion block was more effective in relieving or diminishing pain in patients with malignant disease (73% response) than in patients with benign disease (37% response) (204). Complications of CT-guided celiac plexus block include abdominal or back pain, transient hypotension resulting from loss of visceral vascular tone, and diarrhea (112,204, 317). The hypotension may be severe, and the patient’s blood pressure and heart rate should be closely monitored during and after the procedure (204). No major hemorrhagic or neurologic complications (e.g., paralysis, sexual dysfunction) have been reported with CT-guided ganglion block.

A

B Figure 3-35 CT-guided ganglion block (posterior approach) in a patient with intractable abdominal pain secondary to chronic pancreatitis. A: Using a right posterior paravertebral approach with the patient prone, a 20-gauge needle was advanced into the region of the celiac plexus at the level of the superior mesenteric artery (arrow). Note that the needle passes medial to the right kidney (K) and that its tip is located immediately anterior to the diaphragmatic crus. AO, aorta; V, inferior vena cava. B: CT scan obtained after injection of dilute contrast medium demonstrates diffusion of the contrast material around the superior mesenteric artery. Subsequent injection of 20 mL of absolute alcohol provided marked pain relief.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 125

Interventional Computed Tomography

125

A

B Figure 3-36 CT-guided celiac ganglion block (anterior approach) in a patient with severe abdominal pain secondary to chronic pancreatitis. The patient’s pain was refractory to high-dose narcoticanalgesics, and he was too uncomfortable to lie in a prone position. A: With the patient supine, a 20-gauge needle was advanced into the region of the celiac plexus using an anterior transhepatic approach. There are surgical clips in the gallbladder fossa from previous cholecystectomy. AO, aorta; V, inferior vena cava; c, diaphragmatic crus. B: CT scan demonstrates free retroperitoneal spread of injected contrast medium. Moderate pain relief was achieved after injection of 20 mL of absolute alcohol.

Tumor Ablation In the past several decades there has been increasing interest in minimally invasive techniques to treat a variety of malignancies. Much of the early work centered on chemical ablation of focal hepatic tumors (203). Since then, there has been considerable attention given to thermal techniques. The location of tumors treated has also expanded to include a variety of locations outside the liver, including: lung, kidney, adrenal gland, breast, soft tissues, and bone.

Chemical Ablation Much of the early work with percutaneous tumor ablation centered on the use of alcohol for chemical ablation. Reported experience focused mainly on therapy of hepatocellular carcinoma (26,171,214,219,220,222,312,338,342,344– 348,367), although a few reports also have described its use in treating metastatic liver lesions (216,219,223,313). Most investigators have used alcohol injection for hepatic tumors less than 3 cm in size and fewer than three in number. More recently, some investigators have expanded the indications for this treatment to five or fewer lesions, as well as lesions up to 5 cm in diameter. Contraindications to percutaneous alcohol injection include extrahepatic metastatic disease and irreversible coagulopathy (203). Tumor puncture and ablation is performed by administering absolute alcohol (i.e., 95% ethanol) through a 20or 22-gauge needle. Alcohol is inexpensive and easily applied, and it can kill tumor cells by causing coagulation necrosis, followed by fibrosis and small-vessel thrombosis (203). Although either ultrasound or CT may be used to

guide the procedure, CT guidance has several advantages. Because absolute alcohol has a low CT attenuation value (approximately –240 Hounsfield units [HU]), it can be seen within the lesion on CT. The area of the treated lesion becomes very hypodense, and therefore untreated areas are easily identified. When ultrasound is used for guidance, microbubbles in the injected alcohol produce marked echogenicity with distal acoustic shadowing, which may obscure the lesion and surrounding tissues, particularly late in the injection. The acoustic shadowing makes it difficult to assess whether the alcohol has diffused through the entire lesion. The main advantage of using ultrasound is that real-time visualization of alcohol diffusion and tumor ablation is possible (203). Usually 8 to 10 mL of alcohol are injected per lesion per treatment session, and the injection is repeated 1 to 2 times per week over 2 to 4 weeks. However, the volume and number of injections are modified depending on patient compliance; lesion size, number, and location; and postcontrast CT findings after therapy (345). Successfully treated lesions become necrotic and appear uniformly low attenuation without evidence of enhancement on contrastenhanced CT scans (Fig. 3-37). Rim enhancement, softtissue nodularity, or enhancing areas within the lesion indicate residual tumor. Results of percutaneous ethanol therapy for hepatocellular carcinoma have been encouraging. In one analysis of 77 patients with hepatocellular carcinoma, it was reported that among 50 patients with three or fewer lesions, all of which were treated by percutaneous ethanol injection, the

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 126

126

Chapter 3

A

B

Figure 3-37 Percutaneous CT-guided ethanol treatment of

C

1-, 2-, 3-, and 4-year survival rates were 89%, 74%, 68%, and 60%, respectively (346). Other authors have reported similar results (214,222,338). With respect to treatment of liver metastases, one group of investigators reported successful alcohol ablation of 11 of 21 metastases (223). Nine of the 11 successfully treated lesions in their series were less than 2 cm in diameter. However, to date, there has been no reported long-term follow up of patients with liver metastases treated with alcohol. The most common complications of percutaneous alcohol ablation therapy are fever and pain. These are usually transient and do not require any specific therapy. In order to minimize reflux of the injected alcohol through the needle track into the peritoneal cavity (which may cause abdominal discomfort), the needle is usually left in place for several minutes after the injection (345). In addition, the number of needle passes into the lesion should be minimized to limit reflux of alcohol back through multiple needle tracks (203). Other minor complications that may occur include mild hemorrhage, pneumothorax, or pleural effusion. Serious complications are rare, although there has been a report of

hepatocellular carcinoma. A: Initial CT shows an enhancing mass (arrows) in the right lobe of the liver. Percutaneous biopsy revealed hepatocellular carcinoma. B: A needle has been inserted into the central portion of the lesion for ethanol administration. C: Two-year follow-up contrast-enhanced CT scan at narrow windows shows a successfully treated lesion of uniformly low attenuation without evidence of contrast enhancement.

subcutaneous seeding of tumor after percutaneous ethanol injection treatment of a liver metastasis (123).

Thermal Ablation As an alternative to chemical ablation, tissue destruction may be achieved by applying extreme temperatures. Cell death may be a result of freezing (cryoablation) or incinerating the tissue. Radiofrequency, interstitial laser, and microwave energies have been used to generate heat for the purpose of tissue ablation. Commercially produced cryoablation, microwave ablation, and radiofrequency ablation systems with FDA approval are currently available. Of these, radiofrequency ablation (RFA) has become the most commonly used and has been established as a cost-effective treatment option in selected patients with focal hepatic malignancies (341). The basis for radiofrequency ablation is the generation of a high-frequency oscillating current to agitate local ions within the tissue. This agitation results in ionic friction and, thus, heat. When thermal energy is applied to cells, they undergo predictable reactions and the degree of cellular damage is a function of degree and duration

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 127

Interventional Computed Tomography

of the heat applied (316). Cellular impairment in the liver begins at 45°C with sinusoidal edema and microvascular changes visible histologically (190). Remaining at this temperature for several hours will result in irreversible cell death. At a temperature of 50° to 55°C, cell death will occur in 4 to 6 minutes, and, at 60°C, cell death is nearly instantaneous. Above 100°C, tissue vaporization and charring occur. Vaporized tissue can then act as an insulator, preventing further heat dispersion and limiting further tissue ablation.

RADIOFREQUENCY ABLATION OF THE LIVER Indications Radiofrequency ablation has been used for both primary and secondary liver tumors; indications for RFA continue to evolve as more data is accumulated and the technology continues to advance. RFA is typically reserved for patients with hepatocellular carcinoma (HCC) who might otherwise undergo surgical therapy, either resection or transplantation. This, characteristically, includes patients with early HCC. Early HCC may be classified as a single tumor less than 5 cm in diameter or 3 or fewer lesions measuring less than 3 cm in the setting of Childs A or B cirrhosis (206). Local ablative therapy may be used as an alternative to resection in patients who are not surgical candidates or as a bridge to transplantation. Radiofrequency ablation may also be used to treat focal liver metastases. Most of the work in this arena has centered on metastatic colorectal carcinoma; nonetheless, there are reports of using thermal ablation in the treatment of neuroendocrine, melanoma, and breast carcinoma metastases (70). As with primary hepatocellular carcinoma, most investigators limit the lesion size to 5 cm and the lesion number to three or fewer.

Patient Selection Successful radiofrequency ablation begins with proper patient selection. Before the procedure, a careful review of the clinical history, appropriate laboratory values (liver function tests, platelets, PT, and PTT), and any recent crosssectional imaging study should be performed. Contrastenhanced MRI or multiple-phase CT is necessary in most cases to quantify tumor size, number, and location, each of which plays an important role in patient selection. The “ideal” patient will have three or fewer small lesions (less than 3 cm) surrounded by hepatic parenchyma and at least 1 cm away from any large vessel or adjacent organ. Radiofrequency ablation is limited in the size of lesion that can effectively be treated. Current commercially available devices are capable of creating a zone of coagulation necrosis about 4 to 6 cm in diameter (72,303). To reduce the risk of local recurrence, a 1-cm circumferential rim of normal parenchyma around the lesion should be included in the zone of ablation: this is analogous to a surgical margin.

127

Therefore, with currently available devices, the largest lesion that can be effectively treated with a single burn would measure 3 to 4 cm in diameter. This size limitation, thus far, has been supported in the literature. Livraghi et al. reported complete necrosis in more than 90% of lesions 3 cm or less in diameter (218). In a separate study, the success rate falls to 71% for lesions 3 to 5 cm and 24% for lesions greater than 5 cm (218). Almost as important as size is location. Certain locations are associated with higher rates of local recurrence. Specifically, surface lesions, such as those in subcapsular locations or adjacent to the diaphragm, can be difficult to fully treat percutaneously and may be better approached with laparoscopic guidance. Lesions in perivascular locations also pose a challenge. Flowing blood in vessels greater than 3 mm can disperse heat, creating a “heat sink,” impeding adequate temperature elevation along the tumor margin (225,244). Lu et al. reported only 52% success for treating perivascular lesions compared to 93% success for treating nonperivascular lesions (225). This finding was confirmed histologically, in a separate study, with successful ablation of 47% of lesions in perivascular location and 88% of lesions in nonperivascular locations (226). The location of the lesion may also influence potential procedural complications. Inadvertent thermal injury can occur when treating lesions that abut adjacent organs, such as the gallbladder, colon, or stomach, resulting in cholecystitis, abscess, or bowel perforation. Lesions in these locations, however, do not constitute a contraindication to percutaneous ablation but should be approached with caution. Reports of treating lesions that abut the gastrointestinal tract and gallbladder show similar success and complication rates compared with treating lesions in other locations (51,52). Nonetheless, it may be prudent to displace the adjacent structure from the lesion with percutaneous infusion of sterile water or CO2 immediately before the ablation procedure. Regardless of the imaging modality used for guidance, the lesion must be visible in order to accurately and effectively target it. Most lesions will be visible with either ultrasound or unenhanced CT, and either modality may be used depending on the operator’s preference. Occasionally, lesions may not be identified with either, and, in this circumstance, performing CT after intraarterial administration of ethiodized oil (Ethiodol) is an option (333,420). Ethiodol-enhanced CT takes advantage of Ethiodol’s avidity to HCC. By injecting Ethiodol into the hepatic artery before the ablation procedure (generally 1 to 7 days), the tumor will appear dense on unenhanced CT scan. The tumor will then be readily visible and may be targeted using standard CT guidance technique (Fig. 3-38). This may become unnecessary as ultrasound contrast media become widely available, providing an alternative method for identifying otherwise “invisible” lesions (285). In addition, there is early work demonstrating the feasibility of using MR guidance, but as yet, this is not widely available (231).

5063_Lee_Ch03pp0095-0144 2/17/06 9:41 AM Page 128

128

Chapter 3

A

B

C

D Figure 3-38 Ethiodol-enhanced CT-guided RFA. A: MRI demonstrates a 2.5-cm enhancing lesion in the right hepatic lobe (arrow). Given the patient’s history of chronic hepatitis and elevated alphafetoprotein levels, this was a presumed hepatocellular carcinoma. B: Because the lesion could not be visualized with either ultrasound or unenhanced CT, a hepatic arteriogram was performed. This demonstrated a completely replaced hepatic artery arising from the superior mesenteric artery. A hypervascular mass was identified corresponding to the lesion on MRI (arrow). Ethiodol was injected into the lesion. C: CT scan demonstrates retained Ethiodol within the lesion (arrow). D: The lesion could then be targeted for radiofrequency ablation. (Arrow) radiofrequency probe with tines deployed in the tumor.

Contraindications There are very few contraindications to radiofrequency ablation. Most patients, even with severe medical comorbidities, are able to tolerate RFA under conscious sedation. There are at least two conditions that place the patient at

higher risk for postprocedure complications: uncorrectable coagulopathy and prior biliary surgery. Patients with uncorrectable coagulopathies are at a higher risk for developing a hemorrhagic complication, even with tract ablation, and patients with surgical bilioenteric anastomosis have been shown to be at increased risk for postablation

5063_Lee_Ch03pp0095-0144 2/17/06 9:41 AM Page 129

Interventional Computed Tomography

cholangitis or liver abscess formation. Caution should also be used in treating patients with pacemakers: the current from RFA could potentially interfere with proper pacemaker function. There is very little data on the use of RFA in patients with pacemakers, although patients have been successfully treated with no interference with the operation of the pacemaker (143). Nonetheless, these patients likely require close cardiac monitoring during the procedure.

Technique The technique of CT-guided RFA is similar to the technique for performing a CT-guided biopsy. The patient is placed on the scanner table, and grounding pads are applied to the patient’s thighs. Most ablation procedures may be done with conscious sedation; however, general anesthesia may be required for select patients. A localizing scan is performed and a skin entry site is selected. The appropriate skin site is sterilely prepped and draped. Local anesthetic such as 1% Lidocaine is administered, and a small dermatotome is made with a no. 11 blade scalpel. If using a coaxial system, the introducer needle is placed into the hepatic parenchyma; otherwise, just the insulated needle electrode is used. The needle is advanced using either conventional CT guidance or CTF. The probe is positioned in the center of the lesion for small lesions; larger lesions may require multiple overlapping burns, and the probe should be positioned accordingly. Once the probe is in position, it should be connected to the generator; a burn of the appropriate size is created following the manufacturer’s algorithm. Space limitations

129

under the gantry may be problematic when targeting lesions deep in the hepatic parenchyma, because these typically require long needle electrodes. Flexible probes (RITA, Mountain View, CA) may be helpful in this circumstance (101). Once the ablation is complete, most devices allow for thermocoagulation of the needle tract. This has two advantages. First, tract ablation reduces the risk of bleeding complications, and second, it may reduce the risk of needle tract seeding. The patient may then be transferred to a holding area, where he or she may be monitored with frequent vital sign evaluations and pain assessments for up to 4 hours. At the completion of the recovery, if the patient is awake and the pain is adequately controlled, he may be discharged home. Patients who experience an unexpected complication, have uncontrolled pain following recovery, or live a significant distance from the treating institution may require an overnight admission.

Imaging Follow-Up CT and MRI are the modalities of choice for follow-up imaging following radiofrequency ablation. Some institutions obtain a contrast-enhanced CT within the first week following the ablation procedure to evaluate for residual tumor and immediate complications (121,244,316). Within the first weeks after ablation, the area of ablation will be hypodense and nonenhancing (244,323). A hyperemic rim of hemorrhage and inflammatory cells may be present that may be difficult to distinguish from residual tumor. The patient then is followed at 3 months and every 3 to 6 months thereafter.

A

B Figure 3-39 Percutaneous CT-guided radiofrequency ablation of hepatocellular carcinoma. A: The needle electrodes have been deployed within segment 4 lesion. A 4-cm burn was performed. B: Three-month follow-up MRI. Post-gadolinium T1-weighted MR image through the zone of ablation demonstrates irregular enhancement within the ablation bed (arrow). This was local tumor recurrence.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 130

130

Chapter 3

By 3 months after the ablation, the hyperemic rim has generally resolved. Thickened, nodular enhancement or increasing size of the ablation area is evidence for residual or recurrent tumor (46,212) (Fig. 3-39). On MRI, the zone of coagulation necrosis should be hypointense on T2-weighted sequences and may be hyper- or hypo-intense on T1-weighted sequences. Like CT, a thin rim of peripheral enhancement may be identified, corresponding to a rim of granulation tissue that will resolve over time. Nodular or asymmetric enhancement, especially if it correlates with an area of increased T2 signal, strongly suggests the presence of tumor (128,212).

Results When evaluating the results of radiofrequency ablation in the liver, there are several factors to be considered. The first is the immediate technical success (i.e., the ability to achieve complete coagulative necrosis in a lesion), and the second is the impact on patient survival. There are numerous reports documenting the efficacy of treating small lesions, both primary HCC and metastases, with RFA. Generally, success rates exceed 90% for lesions less than 3 cm in maximal diameter (68,97,218,320,321, 325,398). More modest success rates are reported for medium-sized lesions, which are 3 to 5 cm in diameter. With nodular HCCs of this size, complete necrosis rates are reported to be 71% to 88% (217,325,398). However, for infiltrating lesions, the ability to achieve complete necrosis falls below 50% (217). Once the lesion exceeds 5 cm, the success rates are generally poor, ranging from 23% to 79% (217,242,398). Several different methods have been proposed to improve the success for treating larger lesions. Shankar et al. demonstrated that the combination of alcohol and radiofrequency ablations consistently resulted in larger necrosis volumes than could be achieved with radiofrequency ablation alone. Success rates are reported as high as 87.6% for lesions greater than 3.5 cm by using multiple, overlapping burns (48), and Rossi et al. reported a success rate of 82% for lesions ranging from 3.5 to 8.5 cm when occluding the hepatic artery during the ablation (322). Despite the ability to achieve complete necrosis in a lesion, the patient remains at risk for tumor recurrence. Recurrence may be local, within the ablation bed, or in a new intrahepatic location. Local recurrence is most likely secondary to residual tumor not identified on postablation imaging. In a recent study of ablation sites within explanted livers, residual tumor was identified in 26% of lesions (226). Of these, only 4 of 11 were identified on cross-sectional imaging. More frequently, tumor recurrence occurs in a new location. In one study looking at long-term results of RFA for HCC, new tumors developed outside the ablation site in 14%, 49%, and 81% of patients at 1, 3, and 5 years, respectively (206). These high recurrence rates are reflective of the fact that RFA is a local therapy and does not address the underlying liver disease or primary disease process.

Though the long-term prognosis for patients with HCC remains poor, there is mounting evidence that locoregional treatment with RFA can improve overall survival. In one recent study, overall survival rates for patients treated with RFA were 97% at 1 year, 71% at 3 years, and 48% at 5 years (206). The median survival was 57 months. In this study, there was a statistical significantly different survival rate for class of underlying cirrhosis and multiplicity of tumor. In another recent study evaluating the impact of RFA on patient survival (330,282) patients with early HCC were treated with RFA. Initial complete response was achieved in 70% of patients. Overall survival rates were 87%, 51%, and 27% at 1, 3, and 5 years, respectively. Multivariate analysis did demonstrate that, in addition to underlying Child-Pugh score, complete necrosis of the lesion was an independent predictor of survival. Similarly, treatment with RFA has shown a survival benefit for patients with metastatic colorectal carcinoma. Solbiati et al. report 1-, 3-, and 5-year survival rates of 93%, 69%, and 46%, respectively, for such patients treated with RFA (358). Another study reported overall median survival of 28.9 months after treatment with RFA (19). In this study, lesion size was the most significant predictor of mortality, and CEA value was the most significant predictor of tumor recurrence.

Complications There are several large studies demonstrating an acceptably low complication rate for RFA of hepatic lesions. Overall complication rates range from 7.1% to 10.6% with a 0.3% to 1.4% mortality rate (63,69,113,221). The most frequently reported complications include intraperitoneal hemorrhage, intrahepatic abscess, and injury to adjacent gastrointestinal structures (221). It has also been shown that patients with biliary-enteric anastomoses are at a higher risk for developing postablation cholangitis or hepatic abscess (343). Occasional needle tract seeding has been reported, but this is rare.

TUMOR ABLATION IN OTHER LOCATIONS Lung Recently, much interest has developed in using radiofrequency energy for thermal ablation of both primary and metastatic lung cancers (4,17,165,207,257,388,416). Lung cancer is now the leading cause of cancer-related deaths, and the lungs are the second most common site for metastases (207). For both primary lung tumors and pulmonary metastases, survival benefits have been documented for surgical resection (5,207). Unfortunately, because of associated comorbidities, many of these patients are not candidates for surgical resection.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 131

Interventional Computed Tomography

131

A

B Figure 3-40 Radiofrequency ablation of non-small cell lung carcinoma. A: Radiofrequency probe has been placed in left lung mass. Note alveolar hemorrhage around introducer needle (curved arrow). B: Two-year follow-up CT scan demonstrates that the size of the lesion has remained stable (arrow).

The lung is uniquely suited to RFA because aerated lung acts as an insulator. Therefore, the deposited energy is primarily focused in the tumor. The aerated lung, however, does not allow visualization of many of these lesions with ultrasound, so CT is the primary modality to guide the ablation. When evaluating a patient for pulmonary RFA, it is important to note the size and number of lesions. In general, RFA should be used to treat only when there are three lesions or fewer per lung and each lesion measures less than 3 cm (207). Patients with bilateral disease may be treated, though it is recommended that only one lung be treated per session (362). It is also important to assess lesion location. As in the liver, lesions adjacent to large vessels may not be able to be adequately treated, and lesions adjacent to vital structures (e.g., airways) may result in complications due to inadvertent thermal injury. The procedure for pulmonary RFA is very similar to that used for hepatic RFA. Once the patient is on the CT table, grounding pads are applied to the patient’s thighs. Scout images are obtained, and the needle path is selected. The appropriate skin area is sterilely prepped and draped, and the tissue is anesthetized down to the pleura. Most investigators perform the procedure with conscious sedation (165,207,362,416), though some routinely use general anesthesia (388). The needle electrode is advanced into the center of the lesion using either standard CT guidance or CTF. When a device with deployable tines is used, the tines should be fully deployed and assessed to ensure that the tines are within the lesion and not contacting any adja-

cent vital structures. The ablation is then performed according to manufacturer’s recommendations. When the ablation is complete, the electrode may be removed while thermoablating the tract. Immediately following ablation of pulmonary lesions, a post-treatment scan should be obtained to detect a pneumothorax. There may be peripheral ground glass opacity around the lesion, which is believed to represent alveolar hemorrhage (165) (Fig. 3-40). The lesion should be hypodense, and if contrast is given, there should be no enhancement. Follow-up imaging should be obtained at 1 month, at 3 months, and every 3 months thereafter. Currently, CT is the modality of choice for follow-up imaging, but positron emission tomography (PET) has also been shown to be effective in evaluating the effectiveness of the ablation (173,361). On follow-up CT, there are two primary signs of residual tumor: enhancement and size. Following complete tumor ablation, the lesion should be hypodense and there should be no enhancement following contrast administration. The lesion may, immediately postablation, be slightly larger than on preablation studies. This is because the ablation is generally 0.5 to 1 cm larger than the initial lesion. It should, however, decrease in size at 1 to 3 months and may continue to shrink for up to 1 year. If complete ablation is not achieved, the lesion may initially get smaller but will generally begin to enlarge again at 6 months (165). Early experience with RFA for thoracic malignancies suggests that it is effective when treating lesions 3 cm or

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 132

132

Chapter 3

less in diameter. In one of the larger published series, Yasui et al. (416) treated 99 thoracic malignancies (primary and metastatic) with RFA. The mean diameter of the lesions was 19.5 mm, and 91% showed no growth progression on follow-up imaging. In another study, complete necrosis was achieved in nearly 70% of lesions 3 cm or less but in only 37% of lesions greater than 3 cm (5). Finally, Lee et al. reported complete necrosis in 100% of lesions less than 3 cm but only 23% complete necrosis in larger lesions (199). As yet, no survival data are available in humans. A survival benefit, though, has been shown in animal models (246). A variety of complications for pulmonary RFA have been reported, most of which are minor (17,165,173,207, 257,362,388,416). Patients often will experience lowgrade fever (
39°C) and productive cough after the procedure (207,362). Pneumothorax has been reported in 18% to 38% of procedures with 0% to 33% of these requiring chest tubes (165,173,362,416). Other reported complications include pleural effusion, bronchopleural fistula, abscess, intraparenchymal hemorrhage, hemothorax, skin burn, subcutaneous emphysema, and malignant pleural seeding (165,362,416). As seen in the liver, small microbubbles can form at the ablation site and pass into nearby veins. These bubbles have been identified in the carotid arteries on ultrasound (413), raising the theoretical risk of cerebral infarction. However, in this same study these patients were evaluated with diffusion-weighted imaging with no evidence for cerebral ischemia (413).

Kidney The rate of incidentally detected renal cell carcinomas has been increasing over the past several years. This rise in incidence has been largely attributed to widespread use of cross-sectional imaging (146). When these lesions are small (
3 cm), the natural history is less aggressive than larger lesions leading to widespread use of nephron-sparing surgeries. The survival data for nephron sparing surgery is excellent with 10-year survival rates reported at about 75% (85, 284). However, not all patients are candidates for surgical resection: some patients are excluded due to high surgical risk from severe medical comorbidities. Other patients may not tolerate surgical resection, because of either limited renal reserve or bilateral disease. In 1997, Zlotta et al. first reported the use of RFA in treating renal tumors (421). Though the data at this point is limited, early series show promise. Renal RFA is currently being limited to patients who are not surgical candidates. Other patients who may benefit from renal RFA are patients with solitary kidney or who have multiple foci of disease (e.g., patients with von Hippel-Lindau syndrome). As in other locations, the two primary predictors for successful ablation of renal tumors are tumor size and tumor location. Just as with treating lesions in the liver and lung, success rates will be better for small lesions (i.e., less than 3 cm in size). In the kidney, though, tumor location is perhaps more important

than size. Tumors may be categorized as exophytic, central, or mixed (103). Lesions that are exophytic are at least partially surrounded by perirenal fat. The fat acts as an insulator, allowing for more effective heat generation within the tumor. In contrast, central lesions frequently are close to vessels in the renal hilum. These vessels can act as a “heat sink,” precluding the generation of adequate temperatures to achieve coagulation necrosis of the entire lesion. In addition, animal models have demonstrated an increased risk for complications when ablating central tumors (200). The procedure is performed in a similar fashion as for treating lesions elsewhere. Ultrasound, CT, laparoscopic, or open approaches have been used successfully. When the ablation is being performed percutaneously, it can generally be performed with conscious sedation; however, some investigators routinely perform renal RFA under general anesthesia. The needle electrode is placed into the center of the lesion and the tines are deployed when using expandable electrodes. It is important to fully visualize the tines and make sure that there is at least a 5-mm distance from the tines to adjacent structures (e.g., colon). When a 5-mm space is not present, the structure may be displaced by percutaneously injecting sterile water or CO2 to reduce the risk of nontarget thermal injury. When the ablation is complete, the tract may be cauterized to reduce the risk of hemorrhage-related complications. Following RFA, the patient is generally followed with either multiphase contrast-enhanced CT or MRI. On CT, the lesion will appear hypodense on all phases, and there should be no enhancement following contrast administration (238) (Fig. 3-41). Any region within the tumor that increases by more than 10 HU following contrast administration should be viewed with suspicion for recurrent or residual disease (Fig. 3-42). In contrast to liver and lung lesions, renal tumors do not tend to decrease in size. Treated exophytic lesions may develop fatty infiltration between the renal parenchyma and the ablated tumor (238). Early work with renal RFA has shown promising results. When defining procedural success by lack of enhancement on postablation imaging studies, the success rates for lesions less than 3 cm often exceeds 90% (84,102,241,324). However, the accuracy of CT in assessing tumor viability has been questioned. In a study by Rendon et al., residual tumor was identified histologically in 3/5 tumors that had been treated with RFA. Of these three tumors, only one had been detected on contrast-enhanced CT (314). Other studies have used histologic analysis to assess efficacy of ablation. Two studies evaluated for tumor viability by staining the tumor specimen for nicotinamide adenine dinucleotide (NADH). In one study, no metabolic activity was identified in 8 of 10 treated lesions (237), and in the other study histologic analysis demonstrated viable tumor in 20 of 20 specimens. When five of these specimens were stained for NADH, four of the five were positive (248). In one of the largest series to date, Gervais et al. ablated 42 RCC tumors in 34 patients (102). In this study, several

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 133

Interventional Computed Tomography

133

A

B

Figure 3-41 CT-guided radiofrequency ablation of renal cell carcinoma. A: Two-centimeter enhancing exophytic mass arising from the posterior left kidney. B: With the patient prone, the radiofrequency probe was placed into the lesion with CT guidance (arrow). C: Three-month follow-up contrast-enhanced CT scan demonstrates no enhancement with the ablated lesion.

large (3 cm) lesions were successfully treated. In fact, all exophytic lesions measuring up to 5 cm in diameter were completely ablated. Eleven lesions had a central sinus component, and only 5 of 11 of these tumors were completely treated. Univariate analysis did demonstrate a statistically significant negative predictive value of the presence of a central component. The reported incidence of complications with renal RFA is acceptably low. The most commonly reported complications are hemorrhage, paresthesias, and pain (102,232). These tend to be self-limited and are more common when a trans-psoas approach is used. Other reported complications include ureteral strictures, urine leakage, segmental renal infarct, and perirenal fluid collections, transient creatinine elevation, and urinary tract infection (146,166).

C

There has been one reported case of needle tract seeding (241).

Bone RFA is also used to treat symptomatic bone lesions. Most of the published work has used RFA to treat symptoms related to osteoid osteomas. Osteoid osteoma is a common primary bone tumor characterized by an abnormal arrangement of osteoid and osteoblasts in a matrix of fibrovascular tissue. These lesions characteristically result in nocturnal pain alleviated by aspirin or nonsteroidal antiinflammatory medications (250,305). The pain associated with osteoid osteomas is believed to be due to production of prostaglandins from the nidus of the lesion; therefore,

5063_Lee_Ch03pp0095-0144 2/17/06 9:41 AM Page 134

134

Chapter 3

A

B

Figure 3-42 Residual tumor following renal radiofrequency ablation. A: Axial T1 post-gadolinium MRI with fat saturation demonstrating a 3.5-cm enhancing mass arising from the posterior left kidney (curved arrow). L, Liver; RK, right kidney; LK, left kidney. B: A 4-cm radiofrequency ablation was performed from a posterior approach with the patient positioned prone. RK, right kidney; LK, left kidney. C: Onemonth follow-up CT demonstrates irregular, nodular enhancement along the posterolateral border of the ablated tumor (arrow) indicating residual disease.

C

treatments are aimed at eradication of the nidus, either by resection or ablation (41). Radiographic confirmation of osteoid osteomas before performing RFA is best done with CT. A hypodense nidus surrounded by sclerotic cortical bone is typical. The lesion brightly enhances: a fact that is useful when differentiating this lesion from a Brodie abscess (411). The ablation procedure is generally performed with CT guidance under general anesthesia with the nidus as the target of the ablation (41,305,397) (Fig. 3-43). A projected path, in general, should be perpendicular to the target to reduce the possibility of needle slipping while one is trying to penetrate the bony cortex. Eleven- to 14-gauge bone biopsies systems are often used to traverse the overlying cortical bone; occasionally a bone drill may be necessary. The probe is then placed through the outer cannula and the nidus ablated at 90°C for 4 to 6 minutes (305,412). When the nidus measures larger than 1 cm, multiple needle placements are suggested to reduce the risk of treatment failure (395). Following the

ablation, 2 to 4 mL of 0.5% bupivacaine may be injected into the ablation site to reduce postablation pain (412). Patients typically may be discharged home the same day following recovery from the ablation procedure and may resume normal activities. Rosenthal et al. recommended suspending vigorous sports activities for 3 months after ablation, but this is controversial (319,412). The patient may expect pain in the ablation site that is different from the original pain. This generally resolves by 24 to 48 hours (80). Symptoms that last more than one month are considered persistent symptoms and are considered a treatment failure (41). Thus far, percutaneous RFA of osteoid osteomas has been effective in controlling symptoms related to osteoid osteomas. Primary success rates range from 76% to 100% with most series reporting better than 90% primary success rates (213,318,394,397,412). Patients with persistent or recurrent symptoms may be retreated with reported secondary success rates of 68% to 100% (213, 394, 412). The incidence of complications is very low. Most reported complications

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 135

135

Interventional Computed Tomography

A

B

Figure 3-43 CT-guided radiofrequency ablation of osteoid osteoma in a 15-year-old male with a 3-year history of right hip pain relieved by nonsteroidal anti-inflammatory medications. A: Right hip radiograph demonstrates an area of sclerosis along the medial margin of the proximal right femur (arrow). B: CT scan through the right hip demonstrates an osteoid osteoma with a nidus of less than 1 cm. C: The radiofrequency probe was positioned in the center of the nidus (arrow). The nidus was ablated at 95°C for 5 minutes. The patient remains asymptomatic at 3-year follow-up.

are minor and consist of thermal injury to the skin or soft tissues. In addition to being used for symptomatic osteoid osteomas, RFA has been used for palliation of painful skeletal metastasis in patients who do not respond to conventional therapies (40,120). In one of the largest series to date (120), 43 patients who either failed or refused conventional palliative treatments underwent RFA of the symptomatic metastasis. With a median 16-week follow up, 95% of patients had a clinically significant reduction in pain. In addi-

C

tion, significant reductions in narcotic requirements were observed in this study.

REFERENCES 1. Abbitt PL, Armstrong P. Percutaneous catheter drainage of necrotic tumors. CT demonstration. J Comput Assist Tomogr. 1989;13: 437–439. 2. Adler OB, Rosenberger A, Peleg H. Fine-needle aspiration biopsy of mediastinal masses. Evaluation of 136 experiences. AJR Am J Roentgenol. 1983;140:893–896.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 136

136

Chapter 3

3. Aeder MI, Wellman JL, Haaga JR, et al. Role of surgical and percutaneous drainage in the treatment of abdominal abscesses. Arch Surg. 1983;118:273–280. 4. Ahrar K, Price RE, Wallace MJ, et al. Percutaneous radiofrequency ablation of lung tumors in a large animal model. JVIR. 2003;14:1037–1043. 5. Akeboshi M, Yamakado K, Nakatsuka A, et al. Percutaneous radiofrequency ablation of lung neoplasms: intial therapeutic response. JVIR. 2004;15:463–470. 6. Allison DJ, Adam A. Percutaneous liver biopsy and tract embolization with steel coils. Radiology. 1988;169:261–263. 7. Alspaugh JP, Bernardino ME, Sewell CW, et al. CT directed hepatic biopsies: increased diagnostic accuracy with low patient risk. J Comput Assist Tomogr. 1983;7:1012–1017. 8. Alterbarmakian VK, Guthinger WP, Yakub YN, et al. Percutaneous kidney biopsies: complications and their management. Urology. 1981;28:118–122. 9. Andriole JG, Haaga JR, Adams AB, et al. Biopsy needle characteristics assessed in the laboratory. Radiology. 1983;148:659–662. 10. Anzai Y, Desalles AAF, Black KL, et al. Interventional MR imaging. Radiographics. 1993;13:897–904. 11. Aronberg DJ, Sagel SS, Jost RG, et al. Percutaneous drainage of lung abscess. AJR Am J Roentgenol. 1979;132:282–283. 12. Attar B, Levendoglu H, Cuasay NS. CT-guided percutaneous aspiration and catheter drainage of pyogenic liver abscesses. Am J Gastroenterol. 1986;81:550–555. 13. Austin JHM, Chohen MB. Value of having a cytopathologist present during percutaneous fine-needle aspiration biopsy of lung: report of 55 cancer patients and metaanalysis of the literature. AJR Am J Roentgenol. 1993;160:175–177. 14. Axel L. Simple method for performing oblique CT-guided needle biopsies. AJR Am J Roentgenol. 1984;143:341–342. 15. Ball WS Jr, Bissett GS, III, Towbin RB. Percutaneous drainage of chest abscesses in children. Radiology. 1989;171:431–434. 16. Bankoff MS, Belkin BA. Technical note: percutaneous fine needle aspiration biopsy using a modified coaxial technique. Cardiovasc Intervent Radiol. 1989;12:43–44. 17. Belfiare G, Moggio G, Tedeschi E, et al. CT-guided radiofreqency ablation: a potential complementary therapy for patients with unresectable primary lung cancer—A preliminary report of 33 patients. AJR Am J Roentgenol. 2004;183:1003–1011. 18. Bennett JD, Kozak RI, Taylor BM, et al. Deep pelvic abscesses: transrectal drainage with radiologic guidance. Radiology. 1992;185: 825–828. 19. Berber E, Pelley R, Siperstein AE. Predictors of survival after radiofrequency thermal ablation of colorectal cancer metastases to the liver: a prospective study. J Clin Oncol. 2005;23:1358–1364. 20. Berkman WA, Bernardino ME, Sewell CW, et al. The computed tomography-guided adrenal biopsy. An alternative to surgery in adrenal mass diagnosis. Cancer. 1984;53:2098–2103. 21. Bernardino ME. Percutaneous biopsy. AJR Am J Roentgenol. 1984; 142:41–45. 22. Bernardino ME. Automated biopsy devices: significance and safety. Radiology. 1990;176:615–616. 23. Bernardino ME, Amerson JR. Percutaneous gastrocystostomy: a new approach to pancreatic pseudocyst drainage. AJR Am J Roentgenol. 1984;143:1096–1097. 24. Bernardino ME, Fernandez M, Neylan J, et al. Pancreatic transplants: CT-guided biopsy. Radiology. 1990;177:709–711. 25. Bernardino ME, Walther MM, Phillips VM, et al. CT-guided adrenal biopsy: accuracy, safety, and indications. AJR Am J Roentgenol. 1985;144:67–69. 26. Boland GW, Lee MJ, Dawson SL, et al. Percutaneous injection of ethanol in a patient with a solitary hepatocellular carcinoma. AJR Am J Roentgenol. 1993;161:1071–1077. 27. Brandt KR, Charboneau JW, Stephens DH, et al. CT- and USguided biopsy of the pancreas. Radiology. 1993;187:99–104. 28. Bressler EL, Kirkham JA. Mediastinal masses: alternative approaches to CT-guided needle biopsy. Radiology. 1994;191: 391–396. 29. Bret PM, Fond A, Bretagnolle M, et al. Percutaneous aspiration and drainage of hydatid cysts in the liver. Radiology. 1988;168: 617–620. 30. Bret PM, Fond A, Casola G, et al. Abdominal lesions: A prospective study of clinical efficacy of percutaneous fine-needle biopsy. Radiology. 1986;159:345–346.

31. Bret PM, Sente J-M, Bretagnolle M, et al. Ultrasonically guided fine-needle biopsy in focal intrahepatic lesions: six years’ experience. J Can Assoc Radiol. 1986;37:5–8. 32. Brolin RE, Nosher JL, Leiman S, et al. Percutaneous catheter versus open surgical drainage in the treatment of abdominal abscesses. Am Surg. 1984;50:102–108. 33. Brown KT, Fulbright RK, Avitablile AM, et al. Cytologic analysis in fine-needle aspiration biopsy: smears vs cell blocks. AJR Am J Roentgenol. 1993;161:629–631. 34. Buecker A, Adam GB, Neuerburg JM, et al. Simultaneous realtime visualization of the catheter tip and vascular anatomy for MR-guided PTA of iliac arteries in an animal model. J Magn Imaging. 2002;16:201–208. 35. Buecker A, Neuerburg JM, Adam GB, et al. Real-time MR fluoroscopy for MR-guided iliac artery stent placement. J Magn Imaging. 2000;12:616–622. 36. Burbank F, Kaye K, Belville J, et al. Image-guided automated core biopsies of the breast, chest, abdomen, and pelvis. Radiology. 1994;191:165–171. 37. Butch RJ, Mueller PR, Ferucci JT Jr, et al. Drainage of pelvic abscesses through the greater sciatic foramen. Radiology. 1986;158: 487–491. 38. Butch RJ, Wittenberg J, Mueller PR, et al. Presacral masses after abdominoperineal resection for colorectal carcinoma: the need for needle biopsy. AJR Am J Roentgenol. 1985;144:309–312. 39. Buy J-N, Moss AA, Singler RC. CT guided celiac plexus and splanchnic nerve neurolysis. J Comput Assist Tomogr. 1982;6:315–319. 40. Callstrom MR, Charboneau JW, Goetz MP, et al. Painful metastases involving bone: feasibility of percutaneous CT- and USguided radio-frequency ablation. Radiology. 2002;224:87–97. 41. Cantwell CP, Obyrne J, Eustace S. Current trends in the treatment of osteoid osteoma with an emphasis on radiofrequency ablation. Eur Radiol.- 2004;14:607–617. 42. Carlson SK, Bender CE, Classic KL, et al. Benefits and safety of CT fluoroscopy in interventional radiologic procedures. Radiology. 2001;219:515–520. 43. Casola G, Nicolet V, van Sonnenberg E, et al. Unsuspected pheochromocytoma: risk of blood-pressure alterations during percutaneous adrenal biopsy. Radiology. 1986;159:733–735. 44. Casola G, van Sonnenberg E, D’Agostino HB, et al. Percutaneous drainage of tubo-ovarian abscesses. Radiology. 1992;182:399–402. 45. Casola G, Withers C, van Sonnenberg E, et al. Percutaneous cecostomy for decompression of the massively distended cecum. Radiology. 1986;158:793–794. 46. Catalano O, Esposito M, Nunziata A, et al. Multiphase helical CT findings after percutaneous ablation procedures for hepatocellular carcinoma. Abdom Imaging. 2000;25:607–614. 47. Charboneau JW, Reading CC, Welch TJ. CT and sonographically guided needle biopsy: current techniques and new innovations. AJR Am J Roentgenol. 1990;154:1–10. 48. Chen M-H, Yang W, Yan K, et al. Large liver tumors: protocol for radiofrequency ablation and its clinical application in 110 patients—mathematical model, overlapping mode, and electrode placement process. Radiology. 2004;232:260–271. 49. Chen SM. A single method of limiting needle depth during percutaneous biopsy. Radiology. 1982;143:269. 50. Chezmar JL, Keith LL, Nelson RC, et al. Liver transplant biopsies with a biopsy gun. Radiology. 1991;179:447–448. 51. Choi D, Lim HK, Kim MJ, et al. Therapeutic efficacy and safety of percutaneous radiofrequency ablation of hepatocellular carcinoma abutting the gastrointestinal tract. AJR Am J Roentgenol. 2004;183:1417–1424. 52. Chopra S, Dodd III GD, Chanin MP, et al. Radiofrequency ablation of hepatic tumors adjacent to the gallbaldder: feasibility and safety. AJR Am J Roentgenol. 2003;180:697–701. 53. Chuang VP, Alspaugh JP. Sheath needle for liver biopsy in highrisk patients. Radiology. 1988;166:261–262. 54. Clark RA, Towbin RB. Abscess drainage with CT and ultrasound guidance. Radiol Clin North Am. 1983;21:445–459. 55. Cohan RH, Newman GE, Braun SD, et al. CT assistance for fluoroscopically guided transthoracic needle aspiration biopsy. J Comput Assist Tomogr. 1984;8:1093–1098. 56. Conn HO, Yesner R. A re-evaluation of needle biopsy in the diagnosis of metastatic cancer for the liver. Ann Intern Med. 1963; 59:53–61.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 137

Interventional Computed Tomography 57. Costello P, Gaa J. Clinical assessment of an interventional CT table. Radiology. 1993;189:284–285. 58. Costello P, Onik G, Cosman E. Computed tomographic-guided stereotaxic biopsy of thoracic lesions. J Thorac Imaging. 1987;2: 27–32. 59. Covey AM, Gandhi R, Brody LA, et al. Factors associated with pneumothorax and pneumothorax requiring treatment after percutaneous lung biopsy in 443 consecutive patients. JVIR. 2004; 15:479–483. 60. Crass JR, Simmons RL, Frick MP, et al. Percutaneous decompression of the colon using CT guidance in Ogilvie syndrome. AJR Am J Roentgenol. 1985;144:475–476. 61. Cronan JJ, Amis ES Jr, Dorfman GS. Percutaneous drainage of renal abscesses. AJR Am J Roentgenol. 1984;142:351–354. 62. Crummy AB, McDermott JC, Wojtowycz M. A technique for embolization of biopsy tracts. AJR Am J Roentgenol. 1989;153:67–68. 63. Curley SA, Marra P, Beaty K, et al. Early and late complications after radiofrequency ablation of malignant liver tumors in 608 patients. Ann Surg. 2004;239:450–458. 64. D’Agostino HB, Sanchez RB, Laoide RM, et al. Anterior mediastinal lesions: transsternal biopsy with CT guidance. Radiology. 1993;189:703–705. 65. D’Agostino HB, van Sonnenberg E, Sanchez RB, et al. Treatment of pancreatic pseudocysts with percutaneous drainage and octreotide. Work in progress. Radiology. 1993;187:685–688. 66. Dahnert WF, Hoagland MH, Hamper UM, et al. Fine-needle aspiration biopsy of abdominal lesions: diagnostic yield for different needle tip configurations. Radiology.- 1992;185:263–268. 67. Daly B, Krebs TL, Wong-You-Cheong JJ, et al. Percutaneous abdominal and pelvic interventional procedures using CT fluoroscopy guidance. AJR Am J Roentgenol. 1999;173:637–644. 68. de Baere T, Elias D, Dromain C, et al. Radiofrequency ablation of 100 hepatic metastases with a mean follow-up of more than 1 year. AJR Am J Roentgenol. 2000;175:1619–1625. 69. de Baere T, Risse O, Kuoch V, et al. Adverse events during radiofrequency treatment of 582 hepatic tumors. AJR Am J Roentgenol. 2003;181:695–700. 70. Decadt B, Siriwardena AK. Radiofrequency ablation of liver tumours: systematic review. Lancet Oncol. 2004;5:550–560. 71. DelMaschio A, Vanzulli A, Sironi S, et al. Pancreatic cancer versus chronic pancreatitis: diagnosis with CA 19-9 assessment, US, CT, and CT-guided fine-needle biopsy. Radiology. 1991;178:95–99. 72. Denys AL, de Baere T, Kuoch V, et al. Radio-frequency tissue ablation of the liver: in vivo and ex vivo experiments with four different systems. Eur Radiol. 2003;13:2346–2352. 73. Deyoe LA, Cronan JJ, Lambiase RE, et al. Percutaneous drainage of renal and perirenal abscesses: results in 30 patients. AJR Am J Roentgenol. 1990;155:81–83. 74. Diaz-Buxo JA, Donadio JV. Complications of percutaneous renal biopsy: an analysis of 1,000 consectutive biopsies. Clin Nephrol. 1975;4:223–227. 75. Dickey JE, Haaga JR, Stellato TA, et al. Evaluation of computed tomography guided percutaneous biopsy of the pancreas. Surg Gynecol Obstet. 1986;163:497–503. 76. Do H, Lambiase RE, Deyoe L, et al. Percutaneous drainage of hepatic abscesses: comparison of results in abscesses with and without intrahepatic biliary communication. AJR Am J Roentgenol. 1991;157:1209–1202. 77. Dondelinger RF, Kurdziel J-C, Gathy C. Percutaneous treatment of pyogenic liver abscess: a critical analysis of results. Cardiovasc Intervent Radiol. 1990;13:174–182. 78. Doppmann JL, Krudy AG, Marx SJ, et al. Aspiration of enlarged parathyroid glands for parathyroid hormone assay. Radiology. 1983;148:31–35. 79. Duckwiler G, Lufkin R, Teresi L, et al. Head and neck lesions: MR-guided aspiration biopsy. Radiology. 1989;170:519–522. 80. Dupuy DE, Goldberg SN. Image-guided radiofrequency tumor ablation: challenges and opportunities—part I. JVIR. 2001;12: 1135–1148. 81. Elvin A, Andersson T, Scheibenpflug L, et al. Biopsy of the pancreas with a biopsy gun. Radiology. 1990;176:677–679. 82. Erwin BC, Brynes RK, Chan WC, et al. Percutaneous needle biopsy in the diagnosis and classification of lymphoma. Cancer. 1986;57:1074–1078.

137

83. Fagelman D, Chess Q. Nonaspiration fine-needle cytology of the liver: a new technique for obtaining diagnostic samples. AJR Am J Roentgenol. 1990;155:1217–1219. 84. Farrell MA, Charboneau WJ, DiMarco DS, et al. Imaging-guided radiofrequency ablation of solid renal tumors. AJR Am J Roentgenol. 2003;180:1509–1513. 85. Fergany AF, Hafez KS, Novick AC. Long-term results of nephron sparing surgery for localized renal cell carcinoma: 10-year followup. J Urol. 2000;163:442–445. 86. Ferrucci JT Jr, Wittenberg J. CT biopsy of abdominal tumors: aids for lesion localization. Radiology. 1978;129:739–744. 87. Ferrucci JT Jr, Wittenberg J, Mueller PR, et al. Diagnosis of abdominal malignancy by radiologic fine-needle aspiration biopsy. AJR Am J Roentgenol. 1980;134:323–330. 88. Finch IJ. Preoperative CT-guided percutaneous localization of small masses with a Kopans needle. AJR Am J Roentgenol. 1991;157:179–180. 89. Fink I, Gamsu G, Harter LP. CT-guided aspiration biopsy of the thorax. J Comput Assist Tomogr. 1982;6:958–962. 90. Fish GD, Stanley JH, Miller KS, et al. Post-biopsy pneumothorax: estimating the risk by chest radiography and pulmonary function tests. AJR Am J Roentgenol. 1988;150:71–74. 91. Franseen CC. Aspiration biopsy, with a description of a new type of needle. N Engl J Med. 1941;224:1054–1058. 92. Frederick PR, Brown TH, Miller MH, et al. A light-guidance system to be used for CT-guided biopsy. Radiology. 1985;154:535–536. 93. Freeny PC, Lewis GP, Traverso LW, et al. Infected pancreatic fluid collections: percutaneous catheter drainage. Radiology. 1988;17: 435–441. 94. Gangi A, Kastler B, Arhan J-M, et al. A compact laser beam guidance system for interventional CT. J Comput Assist Tomogr. 1994; 18:326–328. 95. Gardner D, van Sonnenberg E, D’Agostino HB, et al. CT-guided transthoracic needle biopsy. Cardiovasc Intervent Radiol. 1991; 14:17–23. 96. Gatenby RA, Mulhern CB Jr, Broder GJ, et al. Computedtomographic-guided biopsy of small apical and peripheral upper-lobe lung masses. Radiology. 1984;150:591–592. 97. Gazelle GS, Goldberg SN, Solbiati L, et al. Tumor ablation with radiofrequency energy. Radiology. 2000;217:633–646. 98. Gazelle GS, Haaga JR. Guided percutaneous biopsy of intraabdominal lesions. AJR Am J Roentgenol. 1989;153:929–935. 99. Gazelle GS, Haaga JR, Neuhauser D. Hemostatic proteinpolymer sheath: new method to enhance hemostasis at percutaneous biopsy. Radiology. 1990;175:671–674. 100. Gazelle GS, Haaga JR, Stellato TA, et al. Pelvic abscesses: CTguided transrectal drainage. Radiology. 1991;181:49–51. 101. Gebauer B, Gaffke G, Hunerbein M, et al. Flexible applicator systems for radiofrequency ablation (RFA) of hepatic tumors. (abstract). RoFo. 2003;175:1720–1723. 102. Gervais DA, McGovern FJ, Arellano RS, et al. Renal cell carcinoma: clinical experience and technical success with radiofrequency ablation of 42 tumors. Radiology. 2003;226:417–424. 103. Gervais DA, McGovern FJ, Wood BJ, et al. Radio-frequency ablation of renal cell carcinoma: early clinical experience. Radiology. 2000;217:665–672. 104. Gerzof SG. Triangulation: indirect CT guidance for abscess drainage. AJR Am J Roentgenol. 1981;137:1080–1081. 105. Gerzof SG, Johnson WC, Robbins AH, et al. Expanded criteria for percutaneous abscess drainage. Arch Surg. 1985;120:227–232. 106. Gerzof SG, Johnson WC, Robbins AH, et al. Intrahepatic pyogenic abscesses: treatment by percutaneous drainage. Am J Surg. 1985;149:487–494. 107. Gerzof SG, Robbins AH, Birkett DH, et al. Percutaneous catheter drainage of abdominal abscesses guided by ultrasound and computed tomography. AJR Am J Roentgenol. 1979;133:1–8. 108. Gerzof SG, Robbins AH, Johnson WC, et al. Percutaneous catheter drainage of abdominal abscesses. A five-year experience. N Engl J Med. 1981;305:653–657. 109. Gerzof SG, Spira R, Robbins AH. Percutaneous abscess drainage. Semin Roentgenol. 1981;16:62–71. 110. Gianfelice D, Lepanto L, Perreault P, et al. Value of CT fluoroscopy for percutaneous biopsy procedures. J Vasc Interv Radiol. 2000;11:879–884.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 138

138

Chapter 3

111. Gibbons RP, Bush WH Jr, Burnett LL. Needle tract seeding following aspiration of renal cell carcinoma. J Urol. 1977;118:865–867. 112. Gimenez A, Martinez-Noguera A, Donoso L, et al. Percutaneous neurolysis of the celiac plexus via the anterior approach with sonographic guidance. AJR Am J Roentgenol. 1993;161:1061–1063. 113. Giorgio A, Tarantino L, de Stefano G, et al. Complications after percutaneous saline-enhanced radiofrequency ablation of liver tumors: 3-year experience with 336 patients at a single center. AJR Am J Roentgenol. 2005;184:207–211. 114. Glassberg RM, Sussman SK. Life-threatening hemorrhage due to percutaneous transthoracic intervention: importance of the internal mammary artery. AJR Am J Roentgenol. 1990;154:47–49. 115. Glassberg RM, Sussman SK, Glickstein MF. CT anatomy of the internal mammary vessels: importance in planning percutaneous transthoracic procedures. AJR Am J Roentgenol. 1990; 155:397–400. 116. Glazer HS, Weyman PJ, Sagel SS, et al. Nonfunctioning adrenal masses: incidental discovery on computed tomography. AJR Am J Roentgenol. 1982;139:81–85. 117. Gobien RP, Skucas J, Paris BS. CT-assisted fluoroscopically guided aspiration biopsy of central hilar and mediastinal masses. Radiology. 1981;141:443–447. 118. Gobien RP, Stanley JH, Gobien BS, et al. Percutaneous catheter aspiration and drainage of suspected mediastinal abscesses. Radiology. 1984;151:69–71. 119. Gobien RP, Stanley JH, Vujic I, et al. Thoracic biopsy: CT guidance of thin-needle biopsy of pancreatic and other abdominal masses. AJR Am J Roentgenol. 1984;142:827–830. 120. Goetz MP, Callstrom MR, Charboneau JW, et al. Percutaneous image-guided radiofrequency ablation of painful metastases involving bone: a multicenter study. J Clin Oncol. 2004;22: 300–306. 121. Goldberg SN, Gazelle GS, Compton CC, et al. Treatment of intrahepatic malignancy with radiofrequency ablation. Radiologicpathologic correlation. Cancer. 2000;88:2452–2463. 122. Goldstein HM, Zornoza J, Wallace S, et al. Percutaneous fine needle aspiration biopsy of pancreatic and other abdominal masses. Radiology. 1977;123:319–322. 123. Goletti O, DeNegri F, Pucciarelli M, et al. Subcutaneous seeding after percutaneous ethanol injection of liver metastasis. Radiology. 1992;183:785–786. 124. Goralnik CH, O’Connell DM, El Yousef SJ, et al. CT-guided cutting-needle biopsies of selected chest lesions. AJR Am J Roentgenol. 1988;151:903–907. 125. Greene R. Transthoracic needle aspiration biopsy. Philadelphia: WB Saunders, 1982. 126. Greene R, Szyfelbein WM, Isler RJ, et al. Supplementary tissuecore histology from fine-needle transthoracic aspiration biopsy. AJR Am J Roentgenol. 1985;144:787–792. 127. Gronemeyer DHW, Seibel RMM, Plassmann J. Combination of CT and fluoroscopy workstation for 3-D image processing. London: Blackwell Science, 1990. 128. Guan YS, Sun L, Zhou X-P, et al. Hepatocellular carcinoma treated with interventional procedures: CT and MRI follow-up. World J Gastroenterol. 2004;10:3543–3548. 129. Gupta S, Krishnamurthy S, Broemeling LD, et al. Small (
2-cm) subpleural pulmonary lesions: short- versus long-needle-path CT-guided biopsy—comparison of diagnostic yields and complications. Radiology. 2005;234:631–637. 130. Haaga JR. New techniques for CT-guided biopsies. AJR Am J Roentgenol. 1979;133:633–641. 131. Haaga JR. Interventional CT-guided procedures, 3rd ed. St. Louis: Mosby-Year Book, 1994. 132. Haaga JR, Alfidi RJ. Precise biopsy localization by computed tomography. Radiology. 1976;118:603–607. 133. Haaga JR, Beale SM. Use of CO2 to move structures as an aid to percutaneous procedures. Radiology. 1986;161:829–830. 134. Haaga JR, Kori SH, Eastwood DW, et al. Improved technique for CT-guided celiac ganglia block. AJR Am J Roentgenol. 1984;142: 1201–1204. 135. Haaga JR, Reich NE, Havrilla TR, et al. Interventional CT scanning. Radiol Clin North Am. 1977;15:449–456. 136. Haaga JR, Vanek J. Computed tomographic guided liver biopsy using the Menghini needle. Radiology. 1979;133:405–408.

137. Haaga JR, Weinstein AJ. CT-guided percutaneous aspiration and drainage of abscesses. AJR Am J Roentgenol. 1980;135:1187–1194. 138. Haaga JR, Zelch MG, Alfidi RJ, et al. CT-guided antegrade pyelography and percutaneous nephrostomy. AJR Am J Roentgenol. 1977;128:621–624. 139. Hajek PC, Gylys-Morin VM, Stavas J, et al. Localization grid for MR-guided biopsy. Radiology. 1987;163:825–826. 140. Hall-Craggs MA, Lees WR. Fine-needle aspiration biopsy: pancreatic and biliary tumors. AJR Am J Roentgenol. 1986;147:399–403. 141. Haramati LB, Austin JHM. Complications after CT-guided needle biopsy through aerated versus nonaerated lung. Radiology. 1991;181:778. 142. Harter LP, Moss AA, Goldberg HI, et al. CT-guided fine-needle aspirations for diagnosis of benign and malignant disease. AJR Am J Roentgenol. 1983;140:363–367. 143. Hayes DL, Charboneau JW, Lewis BD, et al. Radiofrequency treatment of hepatic neoplasms in patients with permanent pacemakers. Mayo Clin Proc. 2001;76:950–952. 144. Heaston DK, Handel DB, Ashton PR, et al. Narrow gauge needle aspiration of solid adrenal masses. AJR Am J Roentgenol. 1982;138:1143–1148. 145. Heiberg E, Wolverson MK. Ipsilateral decubitus position for percutaneous CT-guided adrenal biopsy. J Comput Assist Tomogr. 1985;9:217–218. 146. Hines-Peralta A, Goldberg SN. Review of radiofrequency ablation for renal cell carcinoma. Clin Cancer Research. 2004;10: 6328s–6334s. 147. Ho C-S, McLoughlin MJ, McHattie JD, et al. Percutaneous fine needle aspiration biopsy of the pancreas following endoscopic retrograde cholangiopancreatography. Radiology. 1977;125:351–353. 148. Ho C-S, Taylor B. Percutaneous transgastric drainage for pancreatic pseudocyst. AJR Am J Roentgenol. 1984;143:623–625. 149. Hopper KD, Abendroth CS, Sturtz KW, et al. Adjustable-depth biopsy device: comparison with a single-depth biopsy gun in cadaveric organs. AJR Am J Roentgenol. 1993;161:1307–1308. 150. Hopper KD, Abendroth CS, Sturtz KW, et al. Fine-needle aspiration biopsy for cytopathologic analysis: utility of syringe handles, automated guns, and the nonsuction method. Radiology. 1992;185:819–824. 151. Hopper KD, Abendroth CS, Sturtz KW, et al. Blinded comparison of biopsy needles and automated devices in vitro: 1. Biopsy of diffuse hepatic disease. AJR Am J Roentgenol. 1993;161:1293– 1297. 152. Hopper KD, Abendroth CS, Sturtz KW, et al. Blinded comparison of biopsy needles and automated devices in vitro: 2. Biopsy of medical renal disease. AJR Am J Roentgenol. 1993;161:1299– 1301. 153. Hopper KD, Abendroth CS, Sturtz KW, et al. Automated biopsy devices: a blinded evaluation. Radiology. 1993;187:653–660. 154. Hopper KD, Baird DE, Reddy VV. Efficacy of automated biopsy guns versus conventional biopsy needles in the pygmy pig. Radiology. 1990;176:671–676. 155. Hopper KD, Yakes WF. The posterior intercostal approach for percutaneous renal procedures: risk of puncturing the lung, spleen, and liver as determined by CT. AJR Am J Roentgenol. 1990;154:115–117. 156. Hueftle MG, Haaga JR. Effect of suction on biopsy sample size. AJR Am J Roentgenol. 1986;147:1014–1016. 157. Irie T, Kajitani M, Itai Y. CT Fluoroscopy—guided intervention: marked reduction of scattered radioation dose to the physician’s hand by use of a lead plate and an improved I-I device. J Vasc Interv Radiol. 2001;12:1417–1421. 158. Isler RJ, Ferrucci JT Jr., Wittenberg J, et al. Tissue core biopsy of abdominal tumors with a 22 gauge cutting needle. AJR Am J Roentgenol. 1981;136:725–728. 159. Jaffe RB, Arata JA Jr, Matlak ME. Percutaneous drainage of traumatic pancreatic pseudocysts in children. AJR Am J Roentgenol. 1989;152:591–595. 160. Jaques P, Mauro M, Safrit H, et al. CT features of intraabdominal abscesses: prediction of successful percutaneous drainage. AJR Am J Roentgenol. 1986;146:1041–1045. 161. Jaques PF, Staab E, Richey W, et al. CT-assisted pelvic and abdominal aspiration biopsies in gynecological malignancy. Radiology. 1978;128:651–655.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 139

Interventional Computed Tomography 162. Jasinski RW. CT-guided suprasternal approach for mediastinal biopsy. J Comput Assist Tomogr. 1992;16:669–670. 163. Jeffrey RB Jr, Federle MP, Tolentino CS. Perappendiceal inflammatory masses: CT-directed management and clinical outcome in 70 patients. Radiology. 1988;167:13–16. 164. Jeffrey RB Jr, Tolentino CS, Federle MP, et al. Percutaneous drainage of periappendiceal abscesses: review of 20 patients. AJR Am J Roentgenol. 1987;149:59–62. 165. Jin GY, Lee JM, Lee YC, et al. Primary and secondary lung malignancies treated with percutaneous radiofrequency ablation: evaluation with follow-up helical CT. AJR Am J Roentgenol. 2004;183:1013–1020. 166. Johnson DB, Solomon SB, Su L-M, et al. Defining the complications of cryoablation and radio frequency ablation of small renal tumors: a multi-institutional review. J Urol. 2004;172:874–877. 167. Johnson RD, Mueller PR, Ferrucci JT Jr., et al. Percutaneous drainage of pyogenic liver abscesses. AJR Am J Roentgenol. 1985;144:463–467. 168. Johnson WC, Gerzof SG, Robbins AH, et al. Treatment of abdominal abscesses. Comparative evaluation of operative drainage versus percutaneous catheter drainage guided by computed tomography or ultrasound. Ann Surg. 1981;194:510–520. 169. Johnsrude IS, Silverman JF, Weaver MD, et al. Rapid cytology to decrease pneumothorax incidence after percutaneous biopsy. AJR Am J Roentgenol. 1985;144:793–794. 170. Jorulf H, Bennett L. Adjustable automated biopsy device. Radiology. 1992;185:897–898. 171. Joseph FB, Baumgarten DA, Bernardino ME. Hepatocellular carcinoma: CT appearance after percutaneous ethanol ablation therapy. Work in progress. Radiology. 1993;186:553–556. 172. Kane NM, Korobkin M, Francis IR, et al. Percutaneous biopsy of left adrenal masses: prevalence of pancreatitis after anterior approach. AJR Am J Roentgenol. 1991;157:777–780. 173. Kang S, Luo R, Liao W, et al. Single group study to evaluate the feasibility and complications of radiofrequency ablation and usefulness of post treatment positron emission tomography in lung tumours. World J Surgical Oncology. 2004;2:30. 174. Karlson KB, Martin EC, Fankuchen EI, et al. Percutaneous drainage of pancreatic pseudocysts and abscesses. Radiology. 1982;142:619–624. 175. Kato R, Katada K, Anno H, et al. Radiation dosimetery at CT fluoroscopy: physician’s hand dose and development of needle holders. Radiology. 1996;201:576–578. 176. Kattapuram SV, Rosenthal DI. Percutaneous biopsy of skeletal lesions. AJR Am J Roentgenol. 1991;157:935–942. 177. Katz RL, Shirkhoda A. Diagnostic approach to incidental adrenal nodules in the cancer patient. Results of a clinical, radiologic, and fine-needle aspiration study. Cancer. 1985;55:1995–2000. 178. Kazerooni EA, Lim FT, Mikhail A, et al. Risk of pneumothorax in CT-guided transthoracic needle aspiration biopsy of the lung. Radiology. 1996;198:371–375. 179. Kee ST, Ganguly A, Daniel BL, et al. MR-guided transjugular intrahepatic portosystemic shunt creation with use of a hybrid radiography/MR system. JVIR. 2005;16:227–234. 180. Kerlan RK Jr, Jeffrey RB Jr, Pogany AC, et al. Abdominal abscess with low-output fistula: sucessful percutaneous drainage. Radiology. 1985;155:73–75. 181. Kerlan RK Jr, Ring EJ. A replaceable sump lumen catheter for abscess drainage. Gastrointest Radiol. 1986;11:112–113. 182. Kinney TB, Lee MJ, Filomena CA, et al. Fine-needle biopsy: prespective comparison of aspiration versus nonaspiration techniques in the abdomen. Radiology. 1993;186:549–552. 183. Klein JS, Schultz S. Interventional chest radiology. Curr Probl Diagn Radiol. 1992;21:219–268. 184. Klein TS, Neal HS. Needle aspiration biopsy: a critical appraisal eight years and 3,267 specimens later. JAMA. 1978;239:36–39. 185. Klose K-C. CT-guided large-bore biopsy: extrapleural injection of saline for safe transpleural access to pulmonary lesions. Cardiovasc Intervent Radiol. 1993;16:259–261. 186. Koenker RM, Mueller PR, van Sonnenberg E. Interventional radiology of the adrenal glands. Semin Roentgenol. 1988;22:314–322. 187. Koutrouvelis PG, Louie A, Lang E, et al. A three-dimensional stereotactic device for computed tomography-guided invasive diagnostic and therapeutic precedures. Invest Radiol. 1993;28:845–847.

139

188. Kreula J. Effect of sampling technique on specimen size in fine needle aspiration biopsy. Invest Radiol. 1990;25:1294–1299. 189. Kreula J, Virkkunen P, Bondestam S. Effect of suction on specimen size in fine-needle aspiration biopsy. Invest Radiol. 1990;25:1175–1181. 190. Kruskal JB, Oliver B, Huertas J-C, et al. Dynamic intrahepatic flow and cellular alterations during radiofrequency ablation of liver tissue in mice. JVIR. 2001;12:1193–1201. 191. Kuligowska E, Keller E, Ferrucci JT Jr. Treatment of pelvic abscesses: value of one-step sonographically guided transrectal needle aspiration and lavage. AJR Am J Roentgenol. 1995;164:201–206. 192. Kuligowska E, Olsen WL. Pancreatic pseudocysts drained through a percutaneous transgastric approach. Radiology. 1985; 154:79–82. 193. Ladd ME, Quick HH, Debatin JF. Interventional MRA and intravascular imaging. J Magn Imaging. 2000;12:534–546. 194. Lahorra JM, Haaga JR, Stellato TA, et al. Safety of intracavitary urokinase with percutaneous abscess drainage. AJR Am J Roentgenol. 1993;160:171–174. 195. Lambiase RE, Cronan JJ, Dorfman GS, et al. Percutaneous drainage of abscesses in patients with Crohn disease. AJR Am J Roentgenol. 1988;150:1043–1045. 196. Lambiase RE, Deyoe L, Cronan JJ, et al. Percutaneous drainage of 335 consecutive abscesses: results of primary drainage with 1year follow up. Radiology. 1992;184:167–179. 197. Lang EK. Renal, perirenal, and pararenal abscesses: percutaneous drainage. Radiology. 1990;174:109–113. 198. Lang EK, Springer RM, Glorioso LW III, et al. Abdominal abscess drainage under radiologic guidance: causes of failure. Radiology. 1986;159:329–336. 199. Lee JM, Jin GY, Goldberg SN, et al. Percutaneous radiofrequency ablation for inoperable non-small cell lung cancer and metastases: preliminary report. Radiology. 2004;230:125–134. 200. Lee JM, Kim SW, Chung GH, et al. Open radio-frequency thermal ablation of renal VX2 tumors in rabbit model using a cooled-tip electrode: feasibility, safety, and effectiveness. Eur J Radiol. 2003;13:1324–1332. 201. Lee KS, Im J-G, Kim Y-H, et al. Treatment of thoracic multiloculated empyemas with intracavitary urokinase: a prospective study. Radiology. 1991;179:771–775. 202. Lee MJ, Mueller PR, Dawson SL, et al. Measurement of tissue carcinoembryonic antigen levels from fine-needle biopsy specimens: technique and clinical usefulness. Radiology. 1992; 184:717–720. 203. Lee MJ, Mueller PR, Dawson SL, et al. Percutaneous ethanol injection for the treatment of hepatic tumors: indications, mechanism of action, technique, and efficacy. AJR Am J Roentgenol. 1995;164:215–220. 204. Lee MJ, Mueller PR, van Sonnenberg E, et al. CT-guided celiac ganglion block with alcohol. AJR Am J Roentgenol. 1993;161: 633–636. 205. Lee MJ, Rattner DW, Legemate DA, et al. Acute complicated pancreatitis: redefining the role of interventional radiology. Radiology. 1992;183:171–174. 206. Lencioni R, Cioni R, Crocetti L, et al. Early-stage hepatocellular carcinoma in patients with cirrhosis: long-term results of percutaneous image-guided radiofrequency ablation. Radiology. 2005; 234:961–967. 207. Lencioni R, Crocetti L, Cioni R, et al. Radiofrequency ablation of lung malignancies: where do we stand? CVIR. 2004;27:581–590. 208. Lenglinger FX, Schwarz CD, Artmann W. Localization of pulmonary nodules before thorascopic surgery: value of percutaneous staining with methylene blue. AJR Am J Roentgenol. 1994; 163:297–300. 209. Lieberman RP, Hafez GR, Crummy AB. Histology from aspiration biopsy: Turner needle experience. AJR Am J Roentgenol. 1982;138:561–564. 210. Lieberman RP, Waldman SD. Celiac plexus neurolysis with the modified transaortic approach. Radiology. 1990;175:274–276. 211. Lightwood R, Reben HA, Way LW. The risk and accuracy of pancreatic biopsy. Am J Surg. 1976;132:189–194. 212. Limanond P, Zimmerman P, Raman SS, et al. Interpretation of CT and MRI after radiofrequency ablation of hepatic malignancies. AJR Am J Roentgenol. 2003;181:1635–1640.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 140

140

Chapter 3

213. Lindner NJ, Ozaki T, Roedl R, et al. Percutaneous radiofrequency ablation in osteoid osteoma. J Bone Joint Surg Br. 2001;83: 391–396. 214. Livraghi T, Bolondi L, Lazzaroni S, et al. Percutaneous ethanol injection in the treatment of hepatocellular carcinoma in cirrhosis. A study on 207 patients. Cancer. 1992;69:925–929. 215. Livraghi T, Damascelli B, Lombardi C, et al. Risk in fine-needle abdominal biopsy. J Clin Ultrasound. 1983;11:77–81. 216. Livraghi T, Festi D, Monti F, et al. US-guided percutaneous alcohol injection of small hepatic and abdominal tumors. Radiology. 1986;161:309–312. 217. Livraghi T, Goldberg SN, Lazzaroni S, et al. Hepatocellular carcinoma: radio-frequency ablation of medium and large lesions. Radiology. 2000;214:761–768. 218. Livraghi T, Goldberg SN, Lazzaroni S, et al. Small hepatocellular carcinoma: treatment with radio-frequency ablation versus ethanol injection. Radiology. 1999;210:655–661. 219. Livraghi T, Lazzaroni S, Pellicano S, et al. Percutaneous ethanol injection of hepatic tumors: single-session therapy with general anesthesia. AJR Am J Roentgenol. 1993;161:1065–1069. 220. Livraghi T, Salmi A, Bolondi L, et al. Small hepatocellular carcinoma: percutaneous alcohol injection—results in 23 patients. Radiology. 1988;168:313–317. 221. Livraghi T, Solbiati L, Meloni MF, et al. Treatment of focal liver tumors with percutaneous radio-frequency ablation: complications encountered in a multicenter study. Radiology. 2003;226: 441–451. 222. Livraghi T, Vettori C. Percutaneous ethanol injection therapy of hepatoma. Cardiovasc Intervent Radiol. 1990;13:146–152. 223. Livraghi T, Vettori C, Lazzaroni S. Liver metastases: results of percutaneous ethanol injection in 14 patients. Radiology. 1991;179: 709–712. 224. Lorenzo RL, Bradford BF, Black J, et al. Lung abscesses in children: diagnostic and therapeutic needle aspiration. Radiology. 1985;157:79–80. 225. Lu DS, Raman SS, Limanond P, et al. Influence of large peritumoral vessels on outcome of radiofrequency ablation of liver tumors. JVIR. 2003;14:1267–1274. 226. Lu DS, Yu NC, Raman SS, et al. Radiofrequency ablation of hepatocellular carconoma: treatment success as defined by histologic examination of the explanted liver. Radiology. 2005;234: 954–960. 227. Lufkin R, Duckwiler G, Spickler E, et al. MR body stereotaxis: an aid for MR-guided biopsies. J Comput Assist Tomogr. 1988;12: 1088–1089. 228. Lufkin R, Layfield L. Coaxial needle system of MR- and CTguided aspiration cytology. J Comput Assist Tomogr. 1989;13: 1105–1107. 229. Lufkin R, Teresi L, Chiu L, et al. A technique for MR-guided needle placement. AJR Am J Roentgenol. 1988;151:193–196. 230. Lufkin R, Teresi L, Hanafee W. New needle for MR-guided needle placement. AJR Am J Roentgenol. 1987;149:380–382. 231. Mahnken AH, Buecker A, Spuentrup E, et al. MR-guided radiofrequency ablation of hepatic malignancies at 1.5 T: initial results. J Magn Reson Imaging. 2004;19:342–348. 232. Mahnken AH, Gunther RW, Tacke J. Radiofrequency ablation of renal tumors. Eur J Radiol. 2004;14:1449–1455. 233. Mahoney MC, Racadio JM, Merhar GL, et al. Safety and efficacy of kidney transplant biopsy: Tru-Cut needle vs sonographically guided biopsy gun. AJR Am J Roentgenol. 1993;160:325–326. 234. Malt RA. Treatment of pancreatic cancer. JAMA. 1983;250: 1433–1437. 235. Martin EC, Karlson KB, Fankuchen EI, et al. Percutaneous drainage of postoperative intraabdominal abscesses. AJR Am J Roentgenol. 1982;138:13–15. 236. Martino CR, Haaga JR, Bryan PJ, et al. CT-guided liver biopsies: eight years’ experience. Radiology. 1984;152:755–757. 237. Matlaga BR, Zagoria RJ, Woodruff RD, et al. Phase II trial of radiofrequency ablation of renal cancer: evaluation of the kill zone. J Urol. 2002;168:2401–2405. 238. Matsumoto ED, Watumull L, Johnson DB, et al. The radiographic evolution of radiofrequency ablated renal tumors. J Urol. 2004;172:45–48. 239. Mauro M, Jaques P. Modified trocar-cannula system for percutaneous pancreatic abscess drainage. Radiology. 1981;139:227–228.

240. Mauro M, Jaques P, Mandell VS, et al. Pelvic abscess drainage by the transrectal catheter approach in men. AJR Am J Roentgenol. 1985;144:477–479. 241. Mayo-Smith WW, Dupuy DE, Parikh PM, et al. Imaging-guided percutaneous radiofrequency ablation of solid renal masses: techniques and outcomes of 38 treatment sessions in 32 consecutive patients. AJR Am J Roentgenol. 2003;180:1503–1508. 242. Mazzaferro V, Battiston C, Perrone S, et al. Radiofrequency ablation of small hepatocellular carcinoma in cirrhotic patients awaiting liver transplantation. A prospective study. Ann Surg. 2004;240:900–909. 243. McCorkell SJ, Niles NL. Fine–needle aspiration of catecholamineproducing adrenal masses: a possibly fatal mistake. AJR Am J Roentgenol. 1985;145:113–114. 244. McGahan JP, Dodd III GD. Radiofrequency ablation of the liver: current status. AJR Am J Roentgenol. 2001;176:3–16. 245. Meranze SG, LeVeen RF, Burke DR, et al. Transesophageal drainage of mediastinal abscesses. Radiology. 1987;165:395–398. 246. Miao Y, Ni Y, Bosmans H, et al. Radiofrequency ablation for eradication of pulmonary tumor in rabbits. J Surg Res. 2001;99:265–271. 247. Miaux Y, Guermazi A, Gossot D, et al. Laser guidance system for CT-guided procedures. Radiology. 1995;194:282–284. 248. Michaels MJ, Rhee HK, Mourtzinos AP, et al. Incomplete tumor destruction using radio frequency interstitial ablation. J Urol. 2002;168:2406–2410. 249. Miller DA, Carrasco CH, Katz RL, et al. Fine needle aspiration biopsy: the role of immediate cytologic assessment. AJR Am J Roentgenol. 1986;147:155–188. 250. Miller SL, Hoffer FA. Malignant and benign bone tumors. Radiol Clin North Am. 2001;39:673–699. 251. Mindell HJ, Korson R. Comparison of 22-gauge Chiba and screw-tipped needles for cytologic diagnosis of malignancy in in-vitro tumor specimens. Radiology. 1990;176:681–682. 252. Mitchell ML, Carney CN. Cytologic criteria for the diagnosis of pancreatic carcinoma. Am J Clin Pathol. 1985;83:171–176. 253. Mitnick JS, Bosniak MA, Megibow AJ, et al. Non-functioning adrenal adenomas discovered incidentally on computed tomography. Radiology. 1983;148:495–499. 254. Mitty HA, Efremidis SC, Yeh H-C. Impact of fine-needle biopsy on management of patients with carcinoma of the pancreas. AJR Am J Roentgenol. 1981;137:1119–1121. 255. Mladinich CRJ, Ackerman N, Berry CR, et al. Evaluation and comparison of automated biopsy devices. Work in progress. Radiology. 1992;184:845–847. 256. Montaudon M, Latrabe V, Pariente A, et al. Factors influencing accuracy of CT-guided percutaneous biopsies of pulmonary lesions. Eur J Radiol. 2004;14:1234–1240. 257. Morrison PR, van Sonnenberg E, Shanker S, et al. Radiofrequency ablation of thoracic lesions: Part 1, experiments in the normal porcine thorax. AJR Am J Roentgenol. 2005;184:375–380. 258. Moss AA. Interventional computed tomography. In Moss AA, Gamsu G, Genant HK, eds. Computed tomography of the body with magnetic resonance imaging,. 2nd ed. Philidelphia: WB Saunders, 1992:1297–1339. 259. Moulton JS, Moore PT. Coaxial percutaneous biopsy technique with automated biopsy devices: value in improving accuracy and negative predictive value. Radiology. 1993;186:515–522. 260. Moulton JS, Moore PT, Mencini RA. Treatment of loculated pleural effusions with transcatheter intracavitary urokinase. AJR Am J Roentgenol. 1989;153:941–945. 261. Mueller PR. Pancreatic biopsy: striving for excellence. Radiology. 1993;187:15–16. 262. Mueller PR, Dawson SL, Ferrucci JT Jr, et al. Hepatic echinococcal cyst: successful percutaneous drainage. Radiology. 1985;155: 627–628. 263. Mueller PR, Ferrucci JT Jr, Simeone JF, et al. Detection and drainage of bilomas: special considerations. AJR Am J Roentgenol. 1983;140:715–720. 264. Mueller PR, Ferrucci JT Jr., Simeone JF, et al. Lesser sac abscesses and fluid collections: drainage by transhepatic approach. Radiology. 1985;155:615–618. 265. Mueller PR, Miketic LM, Simeone JF, et al. Severe acute pancreatitis after percutaneous biopsy of the pancreas. AJR Am J Roentgenol. 1988;151:493–494.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 141

Interventional Computed Tomography 266. Mueller PR, Saini S, Wittenburg J, et al. Sigmoid diverticular abscesses: percutaneous drainage as an adjunct to surgical resection in 24 cases. Radiology. 1987;164:321–325. 267. Mueller PR, Simeone JF, Butch RJ, et al. Percutaneous drainage of subphrenic abscess: a review of 62 patients. AJR Am J Roentgenol. 1986;147:1237–1240. 268. Mueller PR, Stark D, Simeone JF, et al. MR-guided aspiration biopsy: needle design and clinical trials. Radiology. 1986;161: 605–609. 269. Mueller PR, van Sonnenberg E. Interventional radiology in the chest and abdomen. N Engl J Med. 1990;322:1364–1374. 270. Mueller PR, van Sonnenberg E, Ferrucci JT Jr. Percutaneous drainage of 250 abdominal abscesses and fluid collections. Part II: current procedural concepts. Radiology. 1984;151:343–347. 271. Mueller PR, White EM, Glass-Royal M, et al. Infected abdominal tumors: percutaneous catheter drainage. Radiology. 1989;173: 627–629. 272. Murphy FB, Barefield KP, Steinberg HV, et al. CT- or sonographyguided biopsy of the liver in the presence of ascites: frequency of complications. AJR Am J Roentgenol. 1988;151:485–486. 273. Murphy FB, Bernardino ME. Interventional computed tomography. Curr Probl Diagn Radiol. 1988;17:121–154. 274. Murphy FB, Small WC, Wichman RD, et al. CT and chest radiography are equally sensitive in the detection of pneumothorax after CT-guided pulmonary interventional procedures. AJR Am J Roentgenol. 1990;154:45–46. 275. Nadel L, Baumgartner BR, Bernardino ME. Percutaneous renal biopsies: accuracy, safety, and indications. Urol Radiol 1986; 8: 67–71. 276. Nagata Y, Nishidai T, Abe M, et al. Laser projection system for radiotherapy and CT-guided biopsy. J Comput Assist Tomogr. 1990; 14:1046–1048. 277. Nahman BJ, VanAman ME, McLemore WE, et al. Use of the Rotex needle in percutaneous biopsy of pulmonary malignancy. AJR Am J Roentgenol. 1985;145:97–99. 278. Nawfel RD, Judy PF, Silverman SG, et al. Patient and personnel exposure during CT fluoroscopy—guided interventional procedures. Radiology. 2000;216:180–184. 279. Neff CC, Lawson DW. Boerhaave syndrome: interventional radiologic management. AJR Am J Roentgenol. 1985;145:819–820. 280. Neff CC, van Sonnenberg E, Casola G, et al. Diverticular abscess: percutaneous drainage. Radiology. 1987;163:15–18. 281. Niceforo JR, Coughlin BF. Diagnosis of renal cell carcinoma: value of fine-needle aspiration cytology in patients with metastases or contraindications to nephrectomy. AJR Am J Roentgenol. 1993;161:1303–1305. 282. Nordenstrom B. New instruments for biopsy. Radiology. 1975; 117:474–475. 283. Nosher JL, Plafker J. Fine needle aspiration of the liver with ultrasound guidance. Radiology. 1980;136:177–180. 284. Novick AC. Laparoscopic and partial nephrectomy. Clin Cancer Research 2004;10:6322S–6327S. 285. Numata K, Isozaki T, Oazwa Y, et al. Percutaneous ablation therapy guided by contrast-enhanced sonography for patients with hepatocellular carcinoma. AJR Am J Roentgenol. 2003;180:143–149. 286. Nunez D Jr, Yrizarry JM, Russell E, et al. Transgastric drainage of pancreatic fluid collections. AJR Am J Roentgenol. 1985;145: 815–818. 287. Ohno Y, Hatabu H, Takenaka D, et al. CT-guided transthoracic needle aspiration biopsy of small (
20 mm) solitary pulmonary nodules. AJR Am J Roentgenol. 2003;180:1665–1669. 288. Onik G, Cosman ER, Wells TH Jr, et al. CT-guided aspirations for the body: comparison of hand guidance with stereotaxis. Radiology. 1988;166:389–394. 289. Onik G, Costello P, Cosman ER, et al. CT body stereotaxis: an aid for CT-guided biopsies. AJR Am J Roentgenol. 1986;146: 163–168. 290. Oyen RH, Van Poppel HP, Ameye FE, et al. Lymph node staging of localized prostatic carcinoma with CT and CT-guided fineneedle aspiration biopsy: prospective study of 285 patients. Radiology. 1994;190:315–322. 291. Pagani JJ. Biopsy of focal hepatic lesions. Comparison of 18 and 22 gauge needles. Radiology. 1983;147:673–675. 292. Pagani JJ. Normal adrenal glands in small cell lung carcinoma: CT-guided biopsy. AJR Am J Roentgenol. 1983;140:949–951.

141

293. Pagani JJ. Non-small cell lung carcinoma adrenal metastases. Computed tomagraphy and percutaneous needle biopsy in their diagnosis. Cancer. 1984;53:1058–1060. 294. Palestrant AM. Comprehensive approach to CT-guided procedures with a hand-held guidance device. Radiology. 1990;174: 270–272. 295. Papanicolaou N, Mueller PR, Ferrucci JT Jr., et al. Abscess-fistula association: radiologic recognition and percutaneous management. AJR Am J Roentgenol. 1984;143:811–815. 296. Park JK, Kraus FC, Haaga JR. Fluid flow during percutaneous drainage procedures: an in vitro study of the effects of fluid viscosity, catheter sized, and adjunctive urokinase. AJR Am J Roentgenol. 1993;160:165–169. 297. Parker LA, Melton JW, Delany DJ, et al. Percutaneous small bore catheter drainage in the management of lung abscesses. Chest. 1987;92:213–218. 298. Parker SH, Hopper KD, Yakes WF, et al. Image-directed percutaneous biopsies with a biopsy gun. Radiology. 1989;171:663–669. 299. Parker SH, Lovin JD, Jobe WE, et al. Stereotactic breast biopsy with a biopsy gun. Radiology. 1990;176:741–747. 300. Parker SH, Yakes WF, Hopper KD, et al. Technical note: adaptation of the Bard prostate biopsy gun for CT-guided abdominal biopsies. Cardiovasc Intervent Radiol. 1989;12:50–52. 301. Paulson EK, Sheafor DH, Enterline DS, et al. CT fluoroscopyguided interventional procedures: techniques and radiation dose to radiologists. Radiology. 2001;220:161–167. 302. Pelaez JC, Hill MC, Dach JL, et al. Abdominal aspiration biopsies. Sonographic vs computed tomographic guidance. JAMA. 1983;250:2663–2666. 303. Pereira PL, Trubenbach J, Schenk M, et al. Radiofrequency ablation: in vivo comparison of four commercially available devices in pig livers. Radiology. 2004;232:482–490. 304. Phillips VM, Knopf DR, Bernardino ME. Percutaneous hepatic biopsy in suspected pancreatic carcinoma. J Comput Tomogr. 1984;8:307–310. 305. Pinto CH, Taminiau AHM, Vanderschueren GM, et al. Technical considerations in CT-guide radiofrequency thermal ablation of osteoid osteoma: tricks of the trade. AJR Am J Roentgenol. 2002;179:1633–1642. 306. Plunkett MB, Peterson MS, Landreneau RJ, et al. Peripheral pulmonary nodules: preoperative percutaneous needle localization with CT guidance. Radiology. 1992;185:274–276. 307. Pokieser P, Kain R, Helbich T, et al. Renal biopsy: in vitro and in vivo comparison of a new automatic biopsy device and conventional biopsy systems. Work in progress. Radiology. 1993;186:573–576. 308. Poster RB, Jones DB, Spirt BA. Percutaneous pediatric renal biopsy: use of the biopsy gun. Radiology. 1990;176:725–727. 309. Price RB, Bernardino ME, Berkman WA, et al. Biopsy of the right adrenal gland by the transhepatic approach. Radiology. 1983; 148:566. 310. Quinn SF, Demlow T, Dunkley B. Temno biopsy needle: evaluation of efficacy and safety in 165 biopsy procedures. AJR Am J Roentgenol. 1992;158:641–643. 311. Ravinsky E, Morales C. Diagnosis of lymphoma by image-guided needle biopsies: fine needle aspiration biopsy, core biopsy or both? Acta Cytol. 2005;49:51–57. 312. Redvanly RD, Chezmar JL. Percutaneous ethanol ablation therapy of malignant hepatic tumors using CT guidance. Semin Intervent Radiol. 1993;10:82–87. 313. Redvanly RD, Chezmar JL, Strauss RM, et al. Malignant hepatic tumors: safety of high-dose percutaneous ethanol ablation therapy. Radiology. 1993;188:283–285. 314. Rendon RA, Kachura JR, Sweet JM, et al. The uncertainty of radiofrequency treatment of renal cell carcinoma: findings at immediate and delayed nephrectomy. J Urol. 2002;167:1587–1592. 315. Reyes GD. A guidance device for CT-guided procedures. Radiology. 1990;176:863–864. 316. Rhim H, Goldberg SN, Dodd III GD, et al. Essential techniques for successful radiofrequency thermal ablation of malignant hepatic tumors. Radiographics. 2001;21:S17–S39. 317. Romanelli DF, Beckmann CF, Heiss FW. Celiac plexus block: efficacy and safety of the anterior approach. AJR Am J Roentgenol. 1993;160:497–500. 318. Rosenthal DI. Percutaneous radiofrequency treatment of osteoid osteomas. Semin Musculoskelet Radiol. 1997;1:265–272.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 142

142

Chapter 3

319. Rosenthal DI, Springield DS, Gebhardt MC, et al. Osteoid osteoma: percutaneous radio-frequency ablation. Radiology. 1995;197:451–454. 320. Rossi S, Buscarini E, Garbagnati F, et al. Percutaneous treatment of small hepatic tumors by an expandable RF needle electrode. AJR Am J Roentgenol. 1998;170:1015–1022. 321. Rossi S, Di Stasi M, Buscarini E, et al. Percutaneous radiofrequency interstitial thermal ablation in the treatment of small hepatocellular carcinoma. Cancer J Sci Am. 1995;1:73. 322. Rossi S, Garbagnati F, Lencioni R, et al. Percutaneous radiofrequency thermal ablation of nonresectable hepatocellular carcinoma after occlusion of tumor blood supply. Radiology. 2000;217:119–126. 323. Rossi S, Garbagnati F, Rosa L, et al. Radiofrequency thermal ablation for treatment of hepatocellular carcinoma. Int J Clin Oncol. 2002;7:225–235. 324. Roy-Choudhury SH, Cast JEI, Cooksey G, et al. Early experience with percutaneous radiofrequency ablation of small solid renal masses. AJR Am J Roentgenol. 2003;180:1055–1061. 325. Ruzzenente A, de Manzoni G, Molfetta M, et al. Rapid progression of hepatocellular carcinoma after radiofrequency ablation. World J Gastroenterol. 2004;10:1137–1140. 326. Sacks BA, Palestrant A, Vine H, et al. Catheter/needle assembly for drainage of fluid collections. AJR Am J Roentgenol. 1981;137: 418–419. 327. Sacks D, Banner MP, Meranze SG, et al. Renal and related retroperitoneal abscesses: percutaneous drainage. Radiology. 1988; 167:447–451. 328. Sacks D, Robinson ML. Transgastric percutaneous drainage of pancreatic pseudocysts. AJR Am J Roentgenol. 1988;151:303–306. 329. Saini S, Mueller PR, Wittenberg J, et al. Percutaneous drainage of diverticular abscess. An adjunct to surgical therapy. Arch Surg. 1986;121:475–478. 330. Sala M, Llovet JM, Vilana R, et al. Initial response to percutaneous ablation predicts survival in patients with hepatocellular carcinoma. Hepatology. 2004;40:1352–1360. 331. Santos JEC, Leiman G. Nonaspiration fine needle cytology: application of new technique to nodular thyroid disease. Acta Cytol. 1988;32:353–356. 332. Sateriale M, Cronan JJ, Savadier LD. A 5-year experience with 307 CT-guided renal biopsies: results and complications. J Vasc Interv Radiol. 1991;2:401–407. 333. Sato M, Watanabe Y, Tokui K, et al. CT-guided treatment of ultrasonically invisible hepatocellular carcinoma. Am J Gastroenterol. 2000;95:2102–2106. 334. Schultz T, Puccini S, Schneider J-P, et al. Interventional and intraoperative MR: review and update of techniques and clinical experience. Eur Radiol. 2004;14:2212–2227. 335. Schweitzer ME, Deely DM. Percutaneous biopsy of osteolytic lesions: use of a biopsy gun. Radiology. 1993;189:615–616. 336. Seibel RMM, Gronemeyer DHW, Sorensen RAL. Percutaneous nucleotomy with CT and fluoroscopic guidance. J Vasc Interv Radiol. 1992;3:571–576. 337. Seibel RMM, Gronemeyer DHW, Weigand H. Thorax biopsy: lung, pleura, mediastinum. In Seibel RMM, Gronemeyer DHW, eds. Interventional computed tomography. London: Blackwell Science, 1990: 15–30. 338. Seki T, Nonaka T, Kubota Y, et al. Ultrasonically guided percutaneous ethanol injection therapy for hepatocellular carcinoma. Am J Gastroenterol. 1989;84:1400–1407. 339. Shah RM, Spirn PW, Salazar AM, et al. Localization of peripheral pulmonary nodules for thoracoscopic excision: value of CTguided wire placement. AJR Am J Roentgenol. 1993;161: 279–283. 340. Sheafor DH, Paulson EK, Kliewer MA, et al. Comparison of sonographic and CT guidance techniques: does CT fluoroscopy decrease procedure time? AJR Am J Roentgenol. 2000;174: 939–942. 341. Shetty SK, Rosen MP, Raptopoulos V, et al. Cost-effectiveness of percutaneous radiofrequency ablation for malignant hepatic neoplasms. JVIR. 2001;12:823–833. 342. Sheu J-C, Sung J-L, Huang G-T, et al. Intratumor injection of absolute ethanol under ultrasound guidance for the treatment of small hepatocellular carcinoma. Hepatogastroenterology. 1987;34: 255–261.

343. Shibata T, Yamamoto Y, Yamamoto N, et al. Cholangitis and liver abscess after percutaneous ablation therapy for liver tumors: incidence and risk factors. JVIR. 2003;14:1535–1542. 344. Shiina S, Niwa Y, Omata M. Percutaneous ethanol injection therapy for liver neoplasms. Semin Intervent Radiol. 1993;10:57–68. 345. Shiina S, Tagawa K, Unuma T, et al. Percutaneous ethanol injection therapy for the treatment of hepatocellular carcinoma. AJR Am J Roentgenol. 1990;154:947–951. 346. Shiina S, Tagawa K, Unuma T, et al. Percutaneous ethanol injection therapy of hepatocellular carcinoma: analysis of 77 patients. AJR Am J Roentgenol. 1990;155:1221–1226. 347. Shiina S, Tagawa K, Unuma T, et al. Percutaneous ethanol injection therapy for hepatocellular carcinoma. A histopathologic study. Cancer. 1991;68:1524–1530. 348. Shiina S, Yasuda H, Muto H, et al. Percutaneous ethanol injection in the treatment of liver neoplasms. AJR Am J Roentgenol. 1987;149:949–952. 349. Sider L, Davis TM Jr. Hilar masses: evaluation with CT-guided biopsy after negative bronchoscopic examination. Radiology. 1987;164:107–109. 350. Silverman SG, Bloom DA, Seltzer SE, et al. Needle-tip localization during CT-guided abdominal biopsy: comparison of conventional and spiral CT. AJR Am J Roentgenol. 1992;159:1095–1097. 351. Silverman SG, Lee BY, Mueller PR, et al. Impact of positive findings at image-guided biopsy of lymphoma on patient care: evaluation of clinical history, needle size, and pathologic findings on biopsy performance. Radiology. 1994;190:759–764. 352. Silverman SG, Mueller PR, Pinkney LP, et al. Predictive value of image-guided adrenal biopsy: analysis of results of 101 biopsies. Radiology. 1993;187:715–718. 353. Silverman SG, Mueller PR, Saini S, et al. Thoracic empyema: management with image-guided catheter drainage. Radiology. 1988; 169:5–9. 354. Silverman SG, Tuncali K, Adams DF, et al. CT Fluoroscopyguided abdominal interventions: techniques, results, and radiation exposure. Radiology. 1999;212:673–681. 355. Sinner WN. Fine-needle biopsy of hamartomas of the lung. AJR Am J Roentgenol. 1982;138:65–69. 356. Smith EH. The hazards of fine-needle aspiration biopsy. Ultrasound Med Biol. 1984;10:629–634. 357. Smith EH. Complications of percutaneous abdominal fine-needle biopsy. Radiology. 1991;178:253–258. 358. Solbiati L, Livraghi T, Goldberg SN, et al. Percutaneous radiofrequency ablation of hepatic metastases from colorectal cancer: long-term results in 117 patients. Radiology. 2001;221:159–166. 359. Stavas J, van Sonnenberg E, Casola G, et al. Percutaneous drainage of infected and noninfected thoracic fluid collections. J Thorac Imaging. 1987;2:80–87. 360. Steiner E, Mueller PR, Hahn PF, et al. Complicated pancreatic abscesses: problems in interventional management. Radiology. 1988;167:443–446. 361. Steinke K, King J, Glenn DW, et al. Radiologic appearance and complications of percutaneous computed tomography-guided radiofrquency-ablated pulmonary metastases from colorectal carcinoma. J Comput Assist Tomogr. 2003;27:750–757. 362. Steinke K, King J, Glenn DW, et al. Percutaneous radiofrequency ablation of lung tumors with expandable needle electrodes: tips from preliminary experience. AJR Am J Roentgenol. 2004;183: 605–611. 363. Stern EJ, Webb WR, Gamsu G. CT gantry tilt: utility in transthoracic fine-needle aspiration biopsy. Work in progress. Radiology. 1993;187:873–874. 364. Stewart CJR, Coldewey J, Stewart IS. Comparison of fine needle aspiration cytology and needle core biopsy in the diagnosis of radiologically detected abdominal lesions. J Clin Pathol. 2002;55:93–97. 365. Stiles GM, Berne TV, Thommen VD, et al. Fine needle aspiration of pancreatic fluid collections. Am Surg. 1990;56:764–768. 366. Sundaram M, Wolverson MK, Heiberg E, et al. Utility of CTguided abdominal aspiration procedures. AJR Am J Roentgenol. 1982;139:1111–1115. 367. Tanaka K, Okazaki H, Nakamura S, et al. Hepatocellular carcinoma: treatment with a combination therapy of transcatheter arterial embolization and percutaneous ethanol injection. Radiology. 1991;179:713–717. 368. Tatsuta M, Yamamoto R, Yamamura H, et al. Cytologic examination and CEA measurement in aspirated pancreatic material collected by

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 143

Interventional Computed Tomography

369. 370. 371. 372. 373. 374. 375. 376. 377. 378. 379. 380. 381. 382.

383. 384. 385. 386. 387. 388.

389. 390.

391. 392. 393.

percutaneous fine-needle aspiration biopsy under ultrasonic guidance for the diagnosis of pancreatic carcinoma. Cancer. 1983; 52:693–698. Templeton PA, Krasna M. Localization of pulmonary nodules for thoracoscopic resection: use of needle/wire breast-biopsy system. AJR Am J Roentgenol. 1993;160:761–762. Torres WE, Evert MB, Baumgartner BR, et al. Percutaneous aspiration and drainage of pancreatic pseudocysts. AJR Am J Roentgenol. 1986;147:1007–1009. Triller J, Maddern G, Kraft P, et al. CT-guided biopsy of pelvic masses. Cardiovasc Intervent Radiol. 1991;13:63–68. Turner AF, Sargent EN. Percutaneous pulmonary needle biopsy: an improved needle for a simple direct method of diagnosis. AJR Am J Roentgenol. 1968;104:846–850. Tylen U, Arnesjo B, Lindberg LG, et al. Percutaneous biopsy of carcinoma of the pancreas guided by angiography. Surg Gynecol Obstet. 1976;142:737–739. Tyrrel RT, Murphy FB, Bernardino ME. Tubo-ovarian abscesses: CT-guided percutaneous drainage. Radiology. 1990;175:87–89. Ulmer JL, Choplin RH, Reed JC. Image-guided catheter drainage of the infected pleural space. J Thorac Imaging. 1991;6:65–73. Van Allan RJ, Katz MD, Johnson MB, et al. Uncomplicated amebic liver abscess: prospective evaluation of percutaneous therapeutic aspiration. Radiology. 1992;183:827–830. van Sonnenberg E, Casola G, Ho M, et al. Difficult thoracic lesions: CT-guided biopsy experience in 150 cases. Radiology. 1988; 167:457–461. van Sonnenberg E, D’Agostino HB, Casola G, et al. Percutaneous abscess drainage: current concepts. Radiology. 1991;181:617–626. van Sonnenberg E, D’Agostino HB, Casola G, et al. Lung abscess: CT-guided drainage. Radiology. 1991;178:347–351. van Sonnenberg E, Ferrucci JT Jr., Mueller PR, et al. Percutaneous drainage of abscesses and fluid collections: technique, results, and applications. Radiology. 1982;142:1–10. van Sonnenberg E, Hajek PC, Gylys-Morin VM, et al. A wire-sheath system for MR-guided biopsy and drainage: laboratory studies and experience in 10 patients. AJR Am J Roentgenol. 1988; 151:815–817. van Sonnenberg E, Lin AS, Deutsch AL, et al. Percutaneous biopsy of difficult mediastinal, hilar, and pulmonary lesions by computed-tomographic guidance and a modified coaxial technique. Radiology. 1983;148:300–302. van Sonnenberg E, Mueller PR, Ferrucci JT Jr. Percutaneous drainage of 250 abdominal abscesses and fluid collections. Part I: results, failures, and complications. Radiology. 1984;151:337–341. van Sonnenberg E, Mueller PR, Ferrucci JT Jr., et al. Sump catheter for percutaneous abscess and fluid drainage by trocar or Seldinger technique. AJR Am J Roentgenol. 1982;139:613–614. van Sonnenberg E, Mueller PR, Schiffman HR, et al. Intrahepatic amebic abscesses: indications for and results of percutaneous catheter drainage. Radiology. 1985;156:631–635. van Sonnenberg E, Nakamoto SK, Mueller PR, et al. CT- and ultrasound-guided catheter drainage of empyemas after chest-tube failure. Radiology. 1984;151:349–353. van Sonnenberg E, Polansky AD, Wittich GR, et al. Guide wire protection for trocar puncture of fluid collections. AJR Am J Roentgenol. 1985;145:831–832. van Sonnenberg E, Shanker S, Morrison PR, et al. Radiofrequency ablation of thoracic lesions: part 2, initial clinical experience— technical and multidisciplinary considerations in 30 patients. AJR Am J Roentgenol. 2005;184:381–390. van Sonnenberg E, Wing VW, Casola G, et al. Temporizing effect of percutaneous drainage of complicated abscesses in critically ill patients. AJR Am J Roentgenol. 1984;142:821–826. van Sonnenberg E, Wittenberg J, Ferrucci JT Jr., et al. Triangulation method for percutaneous needle guidance: the angled approach to upper abdominal masses. AJR Am J Roentgenol. 1981; 137:757–761. van Sonnenberg E, Wittich GR, Casola G, et al. Complicated pancreatic inflammatory disease: diagnostic and therapeutic role of interventional radiology. Radiology. 1985;155:335–340. van Sonnenberg E, Wittich GR, Casola G, et al. Periappendiceal abscesses: percutaneous drainage. Radiology. 1987;163:23–26. van Sonnenberg E, Wittich GR, Casola G, et al. Percutaneous drainage of infected and noninfected pancreatic pseudocysts: experience in 101 cases. Radiology. 1989;170:757–761.

143

394. Vanderschueren GM, Taminiau AHM, Obermann WR, et al. Osteoid osteoma: clinical results with thermocoagulation. Radiology. 2002;224:82–86. 395. Vanderschueren GM, Taminiau AHM, Obermann WR, et al. Osteoid osteoma: factors for increased risk of unsuccessful thermal coagulation. Radiology. 2004;233:757–762. 396. Vazquez JL, Thorsen MK, Dodds WJ, et al. Evaluation and treatment of intraabdominal bilomas. AJR Am J Roentgenol. 1985; 144:933–938. 397. Venbrux AC, Montague BJ, Murphy KPJ, et al. Image-guided percutaneous radiofrequency ablation for osteoid osteomas. JVIR. 2003;14:375–380. 398. Vivarelli M, Guglielmi A, Ruzzenente A, et al. Surgical resection versus percutaneous radiofrequency ablation in the treatment of hepatocellular carcinoma on cirrhotic liver. Ann Surg. 2004;240: 102–107. 399. von Schreeb T, Arner O, Skovsted G, et al. Renal adenocarcinoma. Is there a risk of spreading tumour cells in diagnostic puncture? Scand J Urol Nephrol. 1967;1:270–276. 400. Wallace MJ, Krishnamurthy S, Broemeling LD, et al. CT-guided percutaneous fine-needle aspiration biopsy of small (
1-cm) pulmonary lesions. Radiology. 2002;225:823–828. 401. Welch TJ, Sheedy PF II, Johnson CD, et al. CT-guided biopsy: prospective analysis of 1,000 procedures. Radiology. 1989;171: 493–496. 402. Welch TJ, Sheedy PF II, Stephens DH, et al. Percutaneous adrenal biopsy: review of a 10-year experience. Radiology. 1994;193:341–344. 403. Wescott JL. Direct percutaneous needle aspiration of localized pulmonary lesions: results in 422 patients. Radiology. 1980; 137:31–35. 404. Wescott JL. Percutaneous catheter drainage of pleural effusions and empyema. AJR Am J Roentgenol. 1985;144:1189–1193. 405. Wescott JL. Transthoracic needle biopsy of the hilum and mediastinum. J Thorac Imaging. 1987;2:41–48. 406. Whelan TV, Healy GF, Patel TG. Renal biopsy: localization using computed tomography. Urol Radiol 1985;7:94–96. 407. White M, Mueller PR, Ferrucci JT Jr., et al. Percutaneous drainage of postoperative abdominal and pelvic lymphoceles. AJR Am J Roentgenol. 1985;145:1065–1069. 408. Whitmire LF, Galambos JT, Phillips VM, et al. Imaging guided percutaneous hepatic biopsy: diagnostic accuracy and safety. J Clin Gastroenterol. 1985;7:511–515. 409. Whittenberg J, Mueller PR, Ferrucci JT Jr, et al. Percutaneous core biopsy of abdominal tumors using 22 gauge needles: further observations. AJR Am J Roentgenol. 1982;139:75–80. 410. Wildermuth S, Debatin JF, Leung DA, et al. MR imaging-guided intravascular procedures: initial demonstration in a pig model. Radiology. 1997;202:578–583. 411. Woertler K. Benign bone tumors and tumor-like lesions: value of cross-sectional imaging. Eur Radiol. 2003;13:1820–1835. 412. Woertler K, Vestring T, Boettner F, et al. Osteoid Osteoma: CTguided percutaneous radiofrequency ablation and follow-up in 47 patients. JVIR. 2001;12:717–722. 413. Yamamoto A, Matsuoka T, Toyoshima M, et al. Assessment of cerebral microembolism during percutaneous radiofrequency ablation of lung tumors using diffusion-weighted imaging. AJR Am J Roentgenol. 2004;183:1785–1789. 414. Yankaskas BC, Staab EV, Craven MB, et al. Delayed complications from fine-needle biopsies of solid masses of the abdomen. Invest Radiol. 1986;21:325–328. 415. Yankelevitz DF, Henschke CI. Needle-tip localization for CTguided biopsies. J Thorac Imaging. 1993;8:241–243. 416. Yasui K, Kanazawa S, Sano Y, et al. Thoracic tumors treated with CT-guided radiofrequency ablation: initial experience. Radiology. 2004;231:850–857. 417. Yeow K-M, Su I-H, Pan K-T, et al. Risk factors of pneumothorax and bleeding. Multivariate analysis of 660 CT-guided coaxial cutting needle lung biopsies. Chest. 2004;126:748–754. 418. Yeow K-M, Tsay P-K, Cheung Y-C, et al. Factors affecting diagnostic accuracy of CT-guided coaxial cutting needle lung biopsy: retrospective analysis of 631 procedures. JVIR. 2003;14: 581–588. 419. Yueh N, Havlvorsen RA Jr., Letourneau JG, et al. Gantry tilt technique for CT-guided biopsy and drainage. J Comput Assist Tomogr. 1989;13:182–184.

5063_Lee_Ch03pp0095-0144 10/13/05 12:04 PM Page 144

144

Chapter 3

420. Zheng XH, Guan YS, Huang J, et al. Detection of hypervascular hepatocellular carcinoma: comparison of multi-dectector CT subtraction angiography and Lipiodol CT. World J Gastroenterol. 2005;11:200–203. 421. Zlotta AR, Wildschutz T, Raviv G, et al. Radiofrequency interstitial ablation (RITA) is a possible new modality for treatment of renal cancer: ex vivo and in vivo experience. J Endourol. 1997;11: 251–258.

422. Zornoza J, Cabanillas FF, Altoff TM, et al. Percutaneous needle biopsy in abdominal lymphoma. AJR Am J Roentgenol. 1981;136:97–103. 423. Zornoza J, Ordonez N, Bernardino ME, et al. Percutaneous biopsy of adrenal tumors. Urology. 1981;18:412–416. 424. Zornoza J, Wallace S, Ordonez N, et al. Fine-needle aspiration biopsy of the liver. AJR Am J Roentgenol. 1980;134:331–334.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 145

4

Neck Franz J. Wippold II

NECK Expanding interest in neck imaging has been prompted by high-resolution rapid computed tomography (CT) and the increased capability of magnetic resonance imaging (MRI). Both of these modalities have proved to be sensitive and reliable in the evaluation of various disease processes (56,120,132,222,231,239,306,314,376,377,447). Coupled with a detailed physical examination and modern endoscopy, imaging has become indispensable in the characterization and staging of neck pathology. CT and MRI provide essential information about the deep extension of clinically detected masses and may delineate additional clinically unsuspected lesions.

TECHNIQUE Computed Tomography CT scanning of the neck for evaluation of subjective symptoms, palpable masses, or known conditions should begin with a general neck survey examination prior to more detailed and focused protocols. Scanning should cover the region from the base of the skull to the clavicles with 4-mmor 5-mm-thick slices. For suspected laryngeal lesions, patients may be scanned with additionally acquired 2-mm slices and a higher zoom factor (304,348). A digital lateral scout radiograph may assist in planning. In patients without significant dental amalgam, scanning can usually be accomplished in a single range, with slices parallel to the laryngeal ventricle (121). In patients with numerous dental

4

The opinions and assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of Defense.

restorations, two ranges may be necessary. The initial range covers anatomy superior to the teeth and the second range angles the gantry to avoid metal artifacts. Instructing the patient to gently lower the shoulders reduces artifacts in the lower neck. Spiral (helical) CT scanning is rapidly replacing conventional dynamic CT scanning (slice-by-slice acquisition) in most medical centers (187). Spiral CT permits rapid scanning of large volumes of tissue during quiet respiration (384). Spiral images are less susceptible to patient motion than conventional CT, although image noise is slightly increased (256,385). Volumetric helical data permit optimal multiplanar and three-dimensional (3D) reconstructions (148,385,462). Moreover, spectacular CT angiograms can be produced with the data (Fig. 4-1). The amount of intravenous contrast may be reduced compared with conventional dynamic or sequential CT (372). The patient is typically scanned using the parameters of 5-mm collimation, 5-mm per second table feed, and image reconstruction at 2-mm to 5-mm increments. Optional thinner collimation and reconstruction intervals, such as 2
2 mm, further enhance anatomic detail and can be used in evaluating known tumors, especially in the larynx (372). Because of the large number of images generated, selective photographic filming may be useful. Performance of spiral CT requires a power injector and at least a 20-gauge venocatheter (170). A typical intravenous contrast injection protocol is 2 mL to 2.5 mL per second for a total volume of 125 mL. An additional 40 mL to 50 mL may be required at the rate of 0.5 mL per second for additional spiral volumes. Scanning is begun 30 seconds after initiation of the injection (384,405). Nonionic contrast is preferred, especially in high-risk, pediatric, or elderly patients in whom the rapid bolus can produce nausea or vomiting (170). As with any power injection technique, care must be taken to avoid extravasation of contrast material.

145

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 146

146

Chapter 4

A

B Figure 4-1 Enhanced spiral computed tomography (CT) scan of mycotic pseudoaneurysm of external carotid artery branch. A: Enhanced spiral CT source image shows bilobed densely enhancing mass (arrowheads) in the left retromandibular masticator space. B: Surface-rendered data demonstrate proximity of the mass (arrows) to the external carotid artery and aid in establishing the diagnosis of pseudoaneurysm.

Dynamic or sequential scanning remains a useful technique for older scanners. Intravenous contrast is administered with a power injector through a venous catheter. Total volume and injection rates of contrast should be tailored to the patient size, venous access, and general medical condition. A typical injection administers a bolus of 50 mL at 2 mL per second, followed by a slow infusion of an additional 90 mL to 130 mL at 0.8 mL per second. An initial delay of 20 seconds to 25 seconds, from the start of the injection to the beginning of scanning, allows adequate intravascular contrast enhancement. An examination requiring two ranges usually requires infusion of more intravenous contrast. The patient is imaged supine while quietly breathing (121,180). Imaging during suspended inspiration may falsely narrow the airway. For patients with chronic obstructive pulmonary disease in whom the glottis may close during expiration to maintain positive airway pressure, slow inspiration during scanning is recommended (307). Hyperextension of the neck in patients with masses should be avoided because of the risk for airway obstruction (231). Patients should be actively encouraged to lay still and avoid coughing or swallowing. Occasionally, the pyriform sinuses do not fully distend, making detection of hypopharyngeal disease difficult. Scanning during a modified Valsalva maneuver, phonation, or with the patient’s head turned away from the nondistended sinus

eliminates this problem (Figs. 4-2 and 4-3) (115,116,294). Reformatted images in the coronal or sagittal plane may be useful (347). Three-dimensional (3D) CT techniques may be useful for radiotherapy planning. Slices are obtained in the conventional transaxial orientation. Using special computer software, selected anatomic structures are then traced and reformatted into a 3D wire diagram depicting the tumor and key adjacent tissues that can then be manipulated to reveal the optimal radiation port. This technique limits extraneous collateral radiation to other organs, such as the salivary glands (95,96). CT is also useful for guiding percutaneous biopsies (118,119,267,457).

Magnetic Resonance Imaging MR scanning of the neck may be accomplished using a variety of techniques. Examinations limited to the suprahyoid region and base of the skull can be performed with a head coil. More extensive survey examinations that include the infrahyoid region are best completed with a neck coil (222). Surface coils improve the signal-to-noise ratio by almost 50% over standard head coils (180). A coronal or sagittal scout sequence is useful for defining scan planes and ranges. The entire neck is scanned with T1-weighted transaxial images using 5-mm-thick slices before and after intravenous contrast administration. A high and low anatomic

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 147

Neck

A

147

B Figure 4-2 Value of modified Valsalva maneuver. A: The right pyriform sinus is not seen on a scan obtained during inspiration; left pyriform sinus (P); laryngeal vestibule (V). B: Image obtained during a modified Valsalva maneuver distends the pyriform sinuses. The aryepiglottic folds (arrows), which separate the laryngeal vestibule from the pyriform sinuses, also are better seen; epiglottis (arrowhead).

range may be necessary to cover the pertinent anatomy. T1-weighted images display anatomic relationships and detect lesions, such as those found in nodes embedded within fat. T1-weighted coronal images are excellent for defining the false cords, true cords, laryngeal ventricle, and floor of the mouth (132,175,180,222,231). T1-weighted sagittal images are useful for evaluating the preepiglottic space and nasopharynx. T2-weighted transaxial images characterize tissue, detect tumor within muscle, demonstrate cysts, and assist differentiation of posttherapy fibrosis from recurrent tumor (130,133). Gradient moment nulling, flow compensation, cardiac gating, and presaturation pulses minimize motion artifacts; phase-encoding gradients should be oriented in the anteroposterior direction to further reduce distracting artifacts (175). Fast spin echo T2-weighted images have become a popular alternative to the conventional spin echo technique. Fast spin echo sequences have relatively short acquisition times (112,216,455,464).

Gadolinium-enhanced images improve the delineation of margins in many lesions (160,174,309,369,419), although lesions embedded in fat may be obscured unless fatsaturation techniques are used (21). Another potential pitfall of enhanced images is that the normal aerodigestive mucosa enhances, therefore possibly obscuring small mucosal tumors (423). Intravenously administered gadolinium agents are typically well tolerated (174,463). The use of double and triple doses of contrast, sometimes valuable in intracranial disease, has little value in neck imaging (422). Fat-suppression techniques improve the conspicuity of soft tissue lesions embedded in adipose tissue by selectively diminishing the hyperintensity of fat on T1-weighted images (21,404). Gadolinium-enhanced, fat-suppressed, T1-weighted images are especially useful in staging nodal disease (Fig. 4-4) (311). Options include the short T1 inversion recovery (STIR) pulse sequence that relies on the short T1 relaxation time of fat to achieve this effect (205). An alternative technique is frequency-selective fat suppression,

A

B Figure 4-3 Value of modified Valsalva maneuver. A: Left aryepiglottic fold and pyriform sinus are not seen. Lymphadenopathy is present in the left internal jugular chain beneath the sternocleidomastoid muscle; right pyriform sinus (P). B: Image obtained during modified Valsalva maneuver clearly demonstrates a carcinoma of the left aryepiglottic fold (arrows) compressing the left pyriform sinus (arrowheads).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 148

148

Chapter 4

Figure 4-4 Metastatic lymphadenopathy. Fat-saturated, enhanced, T1-weighted magnetic resonance image demonstrates extensive left jugular lymphadenopathy (arrowheads).

in which the signal from fat protons is nulled by exploiting the chemical shift differences between lipid and water protons (205). Fat suppression may produce several annoying artifacts. Magnetic susceptibility is more noticeable than with conventional spin echo techniques and may cause signal loss at bone–air interfaces. Inhomogeneous suppression may be noted when comparing the base of the skull with the infrahyoid regions. Failure of suppression may occur at fat– air interfaces, producing an aberrant hyperintense signal (9). The STIR sequence may be misleading because nonfatty tissue with similar T1 values as fat, such as proteinaceous fluid, methemoglobin, and intensely enhancing soft tissue, also may be suppressed. Use of an alternative technique of fat suppression that is independent of T1 relaxation times overcomes this problem (205). Nonenhanced, fat-suppressed, T1-weighted sequences are useful for comparison with enhanced sequences. Fast spin echo T2weighted images also benefit from fat suppression because of the distracting hyperintense fat signal observed with the fast spin echo technique (216,311). Adjunct MR techniques and applications complement standard sequences in assessing neck lesions. The magnetization transfer (MT) technique uses the transfer of magnetization between restricted protons associated with macromolecules and free-water protons to improve the contrast between lesions and background tissue (20,271,445). MT

may be useful in differentiating enhancing lesions from background tissue and in defining unenhanced lesions (124,456). Three-dimensional reconstruction algorithms permit the 3D depiction of the tumor volume embedded within the surrounding soft tissues. This display facilitates the conceptual transformation of two-dimensional information into a 3D format and allows the examiner to “dissect” the display to a desired tissue depth (218,420). New openbore units are being used for guidance of biopsies (213). Yet another emerging technique growing in clinical importance is MR spectroscopy (MRS) (52). MRS measures various tissue metabolites rather than providing just another method for constructing anatomic cross-sectional images. When used in concert with MRI techniques, MRS adds useful metabolic and biochemical correlation (52,209). Although MRS can examine several different species, such as phosphorus-31, fluorine-19, and carbon-13, most investigations have dealt with hydrogen (proton) spectra. Most published applications of MRS have evaluated intracranial processes; however, some investigations have explored the role of MRS in evaluating extracranial head and neck diseases (257,264,265). Production of MR spectra is based on the same principles that are used in imaging. In imaging, frequency encodes spatial position. With MRS, frequency reflects the chemical composition of the tissue scanned. Most discussions of MR physics focus on the generation of the MR signal in isolated homogeneous populations of atomic nuclei. In tissues, however, the actual measured resonant frequencies depend on an additional factor: the surrounding chemical environment. For example, the precession of hydrogen nuclei embedded within muscle and fat is subjected to subtle influences from neighboring atoms such as oxygen, nitrogen, carbon, and even other hydrogen. The observed resonant frequencies of hydrogen within tissue differ slightly from those of an isolated homogeneous population of protons. This phenomenon is known as chemical shift. Depending on the surrounding chemical environment, each compound displays a set of resonant peaks along a frequency range rather than at a single frequency. The measured chemical shift frequency differences compared with isolated hydrogen are generally quite small and are measured in Hertz (Hz). The relative unit of measurement of parts per million (ppm) has been adopted so that spectra may be easily compared at different field strengths. The positions of resonant peaks as graphically depicted on a scale in ppm are characteristic of metabolites in tissues (308). Various marker metabolites identified by MRS may be helpful in studying extracranial head and neck lesions. One such marker, choline (Cho), is a precursor to acetylcholine and phosphatidylcholine. Cho also is involved in cell membrane metabolism. Elevations of Cho spectral peaks suggest increased cellular membrane biosynthesis and cellular proliferation, as seen in several types of tumors. Another metabolite, creatine (Cr), is involved in energy metabolism. Higher values of the measured ratio of the Cho peak to the Cr peak within a variety of tumors compared with adjacent normal tissue have been observed (264,265).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 149

Neck

Used as a tumor marker, MRS could aid in treatment monitoring and the early identification of tumor recurrence. MRS may also prove useful for tumor mapping because the metabolic data could supplement the anatomic information of conventional cross-sectional imaging (257). Although MRS has become an eagerly embraced tool for the study of intracranial processes, the widespread use of MRS for head and neck applications has been limited. First of all, MRS requires a homogeneous magnetic field. Susceptibility artifacts introduced by the paranasal sinuses, airway, and bone and pulsation artifacts from the carotid arteries severely degrade data (257). Second, the large amount of fat in the neck produces a lipid peak that obscures the relatively small Cho and Cr peaks (257). Also, MRS has proved disappointing for identifying precise histology. Therefore, widespread application of this promising technique for head and neck lesions awaits the development of definitive refinements and subsequent criteria for interpreting results.

Comparison of Computed Tomography and Magnetic Resonance Imaging MRI provides anatomic information on the neck that favorably compares with that from CT in most cases; however, its superiority to CT has not been conclusively established for all applications (69,224,412). Some authors enthusiastically endorse MRI as the modality of choice (19,56,80,125,180,222,291,406,415), whereas others are more cautious and favor CT in certain situations, especially since the introduction of the spiral technique (148,356,387, 459,358). A distinct advantage of MRI is its superb soft tissue contrast and multiplanar display (169,222,318,406,447). Blood vessels, masses, and adjacent soft tissues are easily differentiated on MR images. MRI is especially helpful for those patients in whom there is poor distinction on CT images between a mass and the surrounding soft tissue structures of similar density. MRI easily defines lesions in three orthogonal planes without repositioning the patient. CT scanning and, ultimately, anatomic display are primarily limited to the transverse plane. The transversely oriented laryngeal ventricle is seen in only 10% of CT studies, thus hindering the radiologic differentiation of the true and false cords (317). The spiral imaging technique has markedly improved coronal and sagittal reformation capabilities; however, the direct imaging potential of MRI is superior. MRI also better images the lower neck without the degradation from shoulder artifact that occurs with CT. Dental amalgam and densely calcified or ossified cartilages also pose no serious limitations for MR scanning. Intravenous iodinated contrast agents used in CT carry a definite risk of anaphylaxis and of further compromise of marginal renal function (241). The newer nonionic contrast agents may be safer than older compounds for use in patients with previous contrast reactions or impaired renal function (158). In MR scanning, noniodinated gadolinium compounds are safe in individuals who have had previous

149

reactions to contrast material or impaired renal function and are well tolerated by impaired as well as healthy kidneys (161). MRI also has several shortcomings. Motion artifacts from breathing, carotid artery pulsations, and swallowing degrade images (175,441,442). Nondiagnostic examinations occur in almost 20% of patients (54). Moreover, many patients with neck pathology have chronic pulmonary obstructive disease or difficulty controlling secretions and are unable to tolerate these time-consuming examinations (175). A major advantage of CT is the rapid scanning speed, especially with the spiral technique (190,371). Patients with cardiac pacemakers and metallic cochlear implants are restricted from the MR scan because of the effect of the strong magnetic field on these devices (93,288). Strongly ferromagnetic cerebral aneurysm clips and dorsal column stimulators are contraindications to MR scanning because of potential motion and dislodgment (199); newer nonferromagnetic materials have been developed that are not dangerous (31,88,240,272,334,399). Ferromagnetic fragments lodged in sensitive areas, such as the orbit, can be deleterious (193). Iron filings within the eye have caused visual impairment due to the effect of the magnetic field (193). Most patients with orthopedic appliances can be placed into the magnet without any adverse effect; however, an image artifact may be created if the prosthesis is closely located to the scanned body part (193). Dentures and dental amalgam also may create artifacts. The presence of most metal devices within the body can be detected by a screening CT or radiograph prior to scanning, and ferromagnetic sensitivity can be evaluated with magnetometers (102), although this is not always practical in the clinical setting. Other factors influencing the choice of scanning modality are the confining tubular configuration of the magnet used in MR scanners, the length of the examinations, and the often loud noise of the gradient coils. Previously, up to l0% of patients experienced anxiety and claustrophobia (44,103,298), resulting in suboptimal or aborted examinations. More recently, several strategies have been successfully used to alleviate this anxiety and considerably reduce aborted studies. One of the easiest methods is to provide detailed information about the procedure to patients and their families. Forewarning patients about gradient coil noise within the magnet and about the length of the examination can be reassuring. Because no ionizing radiation is involved, families can remain in the scanning room safely during the procedure to encourage the patient. Although of limited value, breathing-relaxation techniques (297) and hypnosis (106) have shown some success. Finally, the uncooperative patient can be sedated when conservative methods fail (17). This procedure should be reserved only for patients in whom the value of an MR study far outweighs the risks of sedation, including cardiovascular and respiratory depression, anaphylaxis, and death (107). Orally administered drugs may have a variable time of onset and duration of effect. The examination may be delayed if the desired sedation is not adequate at

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 150

150

Chapter 4

the time of the scheduled study, adding to the overall cost of the procedure. Personnel must also carefully monitor patients during the study; this may require the presence of additional qualified nursing staff during the procedure. Because of the intense magnetic field surrounding the scanner, ferromagnetic resuscitation equipment may not be permitted into the scanning room should the patient have an untoward reaction to the medication (92). In general, claustrophobia is not an issue with CT. Compared with MRI, the gantry bore of the machine is larger in diameter and the depth of the gantry is shorter. More important, the scanning times are shorter than those for MRI. New open-bore MR scanning systems are appealing to the claustrophobic patient. For the severely debilitated patient, CT may be the most reasonable alternative. In summary, the decision of whether CT or MRI should be used as the primary imaging modality for a neck mass is still controversial and is dependent on the availability of equipment hardware and software, cost, and the experience of the radiologist. CT and MRI have different and often complementary roles. MR images have many advantages, including superior soft tissue image contrast and multiplanar display; however, MRI examinations tend to be lengthy and prone to motion degradation. MRI is excellent in certain circumstances, such as in defining transglottic tumor spread in laryngeal neoplasms; however, CT remains a superbly efficient screening procedure, especially with the spiral technique. MRI is also preferable for lesions located at the skull base, within the masticator, parapharyngeal, and parotid spaces, and within the tongue and floor of the mouth (358). The initial clinical impression and working diagnosis are extremely important in influencing the choice of modality. In patients referred with only rudimentary clinical evaluations or uncertain etiologies, CT is a reasonable initial step. Most adults referred for a neck mass are likely to have metastatic disease, and CT superbly evaluates cervical lymph nodes. Other lesions in special locations, such as the floor of the mouth or base of the skull, are best evaluated with MRI because of its multiplanar display, superior soft tissue contrast, and ability to detect extradural invasion.

ANATOMY Cervical Triangles Prior to the widespread use of cross-sectional imaging, clinicians used a system of surface anatomic landmarks based on easily recognized triangles to localize structures (36,299,300,305,307). The sternocleidomastoid muscle divides the neck into the anterior and posterior cervical triangles (Fig. 4-5). Boundaries of the anterior triangle include the mandible superiorly, the midline anteriorly, and the sternocleidomastoid muscle posteriorly. The hyoid bone further divides the anterior triangle into suprahyoid and infrahyoid portions (305). The suprahyoid portion is

Figure 4-5 Triangles of the neck: submandibular triangle (A); submental triangle (B); carotid triangle (C); muscular triangle (D); occipital triangle (E); subclavian triangle (F). A and B represent the suprahyoid division of the anterior triangle. (Adapted from Reede DL, Whelan MA, Bergeron RT. CT of the soft tissue structures of the neck. Radiol Clin North Am. 1984;22:239–250.)

subdivided into the submandibular and submental triangles by the anterior belly of the digastric muscle. Major components of the submandibular triangle are the submandibular salivary gland and submandibular lymph nodes. The submental triangle is primarily occupied by lymph nodes. The infrahyoid portion of the anterior cervical triangle is subdivided into the carotid triangle and the muscular triangle by the superior belly of the omohyoid muscle. Important structures within the carotid triangle are the common carotid artery, internal jugular vein, vagus nerve, and internal jugular lymph nodes. Finally, the muscular triangle contains the larynx, hypopharynx, trachea, esophagus, thyroid and parathyroid glands, and strap muscles (36,304). The posterior triangle is defined by the sternocleidomastoid muscle anteriorly, the trapezius muscle posteriorly, and the clavicle inferiorly. The inferior belly of the omohyoid muscle further divides the posterior triangle into the occipital triangle and the subclavian triangle. The occipital triangle contains lymph nodes, the spinal accessory nerve, and cutaneous branches of the cervical plexus. Within the subclavian triangle courses the subclavian artery and vein, lymph nodes, portions of the brachial plexus, and the phrenic nerve (36,304).

Cervical Spaces An alternative to the traditional cervical triangle method of organizing the anatomy of the neck is to divide the neck into spaces created by the deep cervical fascia (Fig. 4-6) (36,147,157,162,255,353,354,381,388,438,452). The deep cervical fascia consists of three layers: a superficial investing layer, a middle visceral layer, and a deep perivertebral layer (18,142,438). These fascial layers divide the neck into identifiable compartments. This method is ideally

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 151

Neck

Figure 4-6 Diagram of the spaces of the neck created by the deep cervical fascia. Axial slices through the suprahyoid (A) and infrahyoid (B) regions illustrate cross-sectional relationships of these spaces. Perivertebral space, anterior compartment (1), perivertebral space, posterior compartment (3), retropharyngeal space (3), visceral space including the pharyngeal mucosal space (4), parapharyngeal space and carotid sheath (5), submandibular space (6), sublingual space (7), masticator space (8), parotid space (9), pharyngeal mucosal space (10). (Adapted from Smoker WRK. Normal anatomy of the neck. In: Som PM, Curtin HD, eds. Head and neck imaging, 3rd ed. St. Louis: Mosby Year Book, 1996:711–771.)

suited for the transverse display of anatomy on CT and MR images. Moreover, the differential diagnosis of diseases of the neck is often defined by the space in which the process is found. The cervical spaces of the suprahyoid and infrahyoid neck include the sublingual space, submandibular space, buccal space, parotid space, parapharyngeal space, carotid space, masticator space, pharyngeal mucosal space, visceral space, retropharyngeal space, posterior cervical space, and perivertebral space (Fig. 4-7).

SUBLINGUAL SPACE The paired sublingual spaces are located in the floor of the mouth and are defined by the mandible anteriorly and laterally, the hyoid bone posteriorly, the oral mucosa superiorly, and the mylohyoid muscle inferiorly (438) (see Fig. 4-7). The fan-shaped mylohyoid muscle resembles a gingko leaf or a scallop shell and separates the floor of the mouth from the soft tissues of the neck (Fig. 4-8). The mylohyoid muscle originates along the inner surface of the mandible and inserts on the hyoid bone. The posterior lateral edges of the muscle permit the sublingual space to freely communicate with the submandibular space (67). Dividing the sublingual spaces are the paired midline geniohyoid muscles, which originate on

151

the inferior genial tubercles of the symphysis menti and insert on the hyoid bone, and the paired genioglossus muscles that originate on the superior genial tubercles and form the bulk of the tongue (Fig. 4-9). Separating these muscles sagitally is a midline low-density plane or septum. Lateral to the genioglossus muscles is a lateral low-density plane that contains fat, the submandibular duct, and the sublingual salivary glands. The paired sublingual spaces communicate with each other anteriorly. Also contained within each sublingual space are the hyoglossus muscles, styloglossus muscles, lingual arteries, deep lobes of the submandibular salivary glands, lingual branches of the V3 division of the fifth cranial nerve, glossopharyngeal nerves, and hypoglossal nerves (149,339). The hyoglossus muscle is the surgical landmark separating the superficially located submandibular duct from the lingual artery located deep to the muscle (214). Injury to the hypoglossal nerve usually results in striking atrophy and fatty replacement of the ipsilateral genioglossus muscle within less than 1 month (36). Sublingual space asymmetry indicates pathology, and obliteration of the midline low-density plane or lateral low-density plane by soft tissue attenuation is virtually always the result of malignant neoplasm (214,270). Typical lesions seen in the sublingual space include carcinomas extending from the floor of the mouth and tongue; ranulas, which are retention cysts of the sublingual salivary gland; dermoids and epidermoids; hemangiomas and lymphangiomas; lingual thyroid tissue and thyroglossal duct cysts; abscesses; lymphadenopathy; and calculi within the submandibular duct (206,245,339) (Fig. 4-10). A submandibular duct dilated greater than 3 mm suggests obstruction; often a calculus is identified as a calcific density within the duct on CT (1) (Fig. 4-11). Major and minor salivary gland malignancies may primarily involve the sublingual space or secondarily invade it (Fig. 4-12). Occasionally, schwannomas may arise in this space. Processes that begin in the sublingual space are not confined by any natural barriers and may communicate with the submandibular space and the soft tissues of the neck. MRI is well suited to the study of the sublingual space because of a lack of significant artifacts from dental amalgam and beam hardening and because of its coronal and sagittal slice capabilities (245,409). CT remains an excellent modality for the evaluation of stone disease.

SUBMANDIBULAR SPACE The submandibular space is posterolateral to the sublingual space and contains the superficial lobe of the submandibular salivary gland and lymph nodes (see Fig. 4-7) (67,214,438). The submandibular space communicates freely with the sublingual space because of the lack of limiting fascial boundaries (67,142). Congenital lesions such as cystic hygromas, branchial cleft cysts, dermoids, epidermoids, and thyroglossal duct cysts may occur in the submandibular space. Abscesses are associated with skin thick-

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 152

152

Chapter 4

A

B

Figure 4-7 Transaxial diagrams of the neck through the high suprahyoid (A), low suprahyoid (B), and infrahyoid (C) levels demonstrating the relationships of the sublingual space (SLS); submandibular space (SMS); buccal space (BS); parotid space (PS); prestyloid parapharyngeal space (PrePPS); carotid space (CS), also known as the poststyloid parapharyngeal space (Post PPS) in the suprahyoid neck; masticator space (MS); pharyngeal mucosal space (PMS); the visceral space (VS); the retropharyngeal space (RPS); the posterior cervical space (PCS); the anterior compartment of the perivertebral space (AntPVS); and the posterior compartment of the perivertebral space (Post PVS). (Adapted from Harnsberger HR. The perivertebral space. In: Harnsberger HR. Head and neck imaging, 2nd ed. St. Louis: Mosby Year Book, 1995:105–119 and Smoker WRK. Normal anatomy of the neck. In: Som PM, Curtin HD, eds. Head and neck imaging, 3rd ed. St. Louis: Mosby Year Book, 1996:711–771.)

C

ening, edema of the fat, and gas in over 50% of cases (67). Calculi commonly occur in the submandibular glands. Tumors of the submandibular gland, such as pleomorphic adenomas, mucoepidermoid cancers, and adenocystic cancers, present as soft tissue masses within the gland (Fig. 4-13). The submandibular lymph nodes are important sentinels in the spread of floor of the mouth infections, and malignancies and may be involved with lymphoma (Fig. 4-14).

BUCCAL SPACE The buccal space is a small region anterior to the masseter and lateral to the buccinator muscle (see Fig. 4-7). This space contains the buccal fat pad and is most commonly involved with infection (188). Deeply invasive skin cancers also may involve this space. Dental pathology usually does not involve the buccal space unless the teeth have excep-

tionally long roots. Infections and neoplasms from adjacent spaces, such as the parotid and masticator spaces, also may secondarily involve the buccal space (Fig. 4-15) (41,42).

PAROTID SPACE The parotid space is located posterior to the masseter muscle and is formed by the divided leaves of the superficial cervical fascia (see Fig. 4-7). Contents of the parotid space include the parotid gland, Stenson’s duct, the facial nerve, intraparotid lymph nodes, and blood vessels (36,150,438). Evaluation of masses within the parotid region begins with a determination of the lesion as intraparotid or extraparotid. Lesions are considered intraparotid if 50% or greater of the circumference is surrounded by parotid tissue and the epicenter is lateral to the parapharyngeal space. Intraparotid masses displace the parapharyngeal fat medially and may widen the stylomandibular tunnel (70). Large lesions of the

5063_Lee_Ch04pp0145-0224

11/1/05

3:22 PM

Page 153

Neck

A

153

B Figure 4-8 Diagram of the mylohyoid muscle (M) viewed from below. The muscle originates on the inner surface of the mandible (arrows) and inserts on the hyoid bone (arrowheads). The posterior lateral free edge of the muscle (open arrows) permits communication of the sublingual space with the submandibular space (A). Mylohyoid muscle resembles a Gingko (Gingko biloba) leaf (B).

deep parotid lobe may be difficult to distinguish from primary parapharyngeal masses. Identification of a fat plane between the lesion and the parotid indicates a parapharyngeal space site of origin, whereas direct contiguity of mass to gland indicates a deep parotid lobe origin (Fig. 4-16). Intraparotid lesions must be localized to either the superficial or deep parotid lobes. The facial nerve divides these two lobes. Because the facial nerve is not reliably demonstrated on scans, the retromandibular vein is chosen as an alternative landmark for demarcation (293,402). Superficial lesions may only need a superficial parotidectomy; however, deep lesions usually require a total parotidectomy (402). The margins of the lesion also should be evaluated. Sharply defined margins tend to favor a benign tumor diagnosis, whereas indistinct margins favor a malignant or

inflammatory diagnosis (360,386). Finally, determination of the number of lesions is helpful in suggesting a diagnosis. Pleomorphic adenomas characteristically present as solitary lesions, and metastases and lymphoepithelial cysts are frequently multicentric (150). Differentiating lesions in the parotid gland is facilitated by recalling that the vast majority of tumors in adults are epithelial in origin. Benign epithelial salivary gland adenomas can be divided into pleomorphic and monomorphic varieties. Pleomorphic adenomas, also known as benign mixed tumors, are the most common salivary tumors and make up almost 80% of parotid neoplasms (293,435). These tumors typically arise in the superficial lobe of a middle-aged patient. Lesions more commonly affect women. The tumor consists of epithelial, myoepithelial, and stromal elements

Figure 4-9 Floor of the mouth, normal anatomy. Coronal T1-weighted magnetic resonance image demonstrates the floor of the mouth anatomy. Sublingual space (S); mandible (black asterisk); genioglossus muscles (white asterisks); intrinsic tongue muscles (T); geniohyoid (white dots); mylohyoid muscle (arrowheads); anterior belly of the digastric muscle (white boxes); platysma (arrows). The mylohyoid muscle forms the floor of the mouth. Inferior lateral to this muscle are the submandibular spaces.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 154

154

Chapter 4

Figure 4-10 Tongue carcinoma. A dense mass (arrowheads) extends from the left tongue into the sublingual space. The lesion displaces the genioglossus muscles (arrows) to the right.

(435). Although these tumors are benign, 25% eventually become malignant if untreated (362). Involvement of the facial nerve is an ominous sign of malignant transformation (435). CT usually demonstrates a well-defined, enhancing soft tissue mass. The MR features of pleomorphic adenoma are nonspecific. Imaging is further complicated by the variable appearance of the parotid with age. Glands of children contain less fat than glands of adults (293). Furthermore, intraparotid metastatic lymphadenopathy can mimic a primary salivary lesion (293). Pleomorphic adenomas tend to be well defined, hypointense on T1-weighted images, and markedly hyperintense on T2-weighted images (Fig. 4-17) (150,293, 362,364). Small tumors are usually homogeneous; however, larger tumors can be heterogeneous due to necrosis, fibrosis, hemorrhage, or mucoid matrix (293,364). Calcification may also occur but can be overlooked on MR images. Warthin tumors (also known as papillary cystadenoma lymphomatosum, adenolymphoma, and lymphomatous adenoma) are the second most common lesion in the parotid gland and are considered monomorphic. Warthin tumors lack the mesenchymal stromal component seen in pleomorphic lesions. Papillary structures, mature lymphocytic infiltration, and cystic changes dominate the histology (435). These benign tumors occur almost exclusively in the parotid gland, which is the last salivary gland to encapsulate. A compelling theory of pathogenesis cites the entrapment of heterotopic salivary gland ductal epithelial tissue within parotid lymph nodes (290,362,435). Almost 80% of these lesions are found in men 50 years of age or older (293). Nearly 10% of cases are multicentric, involving ipsilateral or contralateral glands (250,293,435).

Figure 4-11 Computed tomography scan of the floor of the mouth demonstrates a calcified calculus (arrow) obstructing a dilated left submandibular salivary gland duct (arrowheads). The duct courses lateral to the genioglossus muscle (m).

Like pleomorphic adenomas, the imaging appearance of Warthin tumors is nonspecific and variable on CT and MR images. A well-defined parotid mass with smooth margins, cysts, and bilateral or multifocal involvement should suggest the diagnosis (250). On T1-weighted images, focal high signal usually corresponds to cysts rich in proteinaceous fluids and cholesterol crystals. Cysts poor in these components generally have lower signals. On T2-weighted images, an intermediate signal corresponds to the predominately epithelial component and a high signal correlates with cysts or lymphoid proliferation (250). Hemorrhage also may influence signal intensity (386). Malignant parotid lesions are much less common than benign lesions. Although malignant tumors tend to have irregular margins and intermediate hyperintensity on T2-weighted MR images, the differentiation between benign and malignant can be difficult. Facial nerve palsy associated with a parotid lesion favors a malignant diagnosis; however, imaging has limitations when attempting to evaluate the patient with an isolated facial nerve palsy and no obvious mass (186). Mucoepidermoid carcinoma is the most common parotid gland malignancy. This typically large tumor tends to have lobulated borders and moderate hyperintensity.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 155

Neck

A

155

B Figure 4-12 Pleomorphic adenoma, right sublingual salivary gland. A soft tissue mass (asterisk) occupies the right sublingual space lateral to the genioglossus muscles (arrowheads). It is hypointense on a T1-weighted image (A) and hyperintense on a T2-weighted image (B).

Adenocystic carcinoma is the second most common malignancy of the parotid gland and the most common in the submandibular salivary gland. This tumor tends to spread along the facial nerve and other nerves and may recur decades after initial treatment. Adenocystic carcinomas tend to have infiltrative borders and be intermediate in signal intensity on T2-weighted images, although signal characteristics are nonspecific (450). Moreover, MR images may overestimate the extent of tumor infiltration because of an inability to differentiate tumor from edema (341).

In children, 50% of parotid lesions are hemangiomas. Over 75% are present at birth, and 80% occur in females (123,220). Most lesions are of the capillary variety (220). Spontaneous regression may be associated with phlebolith

Figure 4-14 Dental abscess secondarily involving the subFigure 4-13 Pleomorphic adenoma, submandibular salivary gland. Enhanced computed tomography demonstrates a cystic lesion (asterisk) within the right nontender submandibular salivary gland. The surrounding fat is normal.

mandibular space. Computed tomography in a patient with a left mandibular molar abscess reveals a cluster of submandibular lymph nodes (asterisks) surrounding an enlarged, inflamed, and partially necrotic left submandibular salivary gland (arrowheads). The platysma muscle is thickened (arrows) and the subcutaneous fat is edematous.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 156

156

Chapter 4

Figure 4-15 Adenosquamous carcinoma of Stenson’s duct. Computed tomography demonstrates a large, necrotic mass (black arrowheads) within the buccal space anterior to the masseter muscle (m). The mass invades the skin (white arrowheads).

formation. The cavernous variety rarely regresses spontaneously (220). On CT images, lesions are usually homogeneous and of soft tissue density. On MR images, lesions are usually isointense to muscle on T1-weighted images and hyperintense on T2-weighted images (123). Associated enlarged vessels may be seen, aiding in differentiation from lymphangiomas. The parotid gland contains lymph nodes and can therefore be involved with metastatic disease. A sometimes overlooked primary site of consideration is skin carcinoma and cutaneous melanoma. Patients and referring physicians

alike may dismiss a past resected lesion. Therefore, the radiologist should pursue the issue of a possible cutaneous malignancy when confronted with a metastatic parotid lymph node (97). Other lesions occurring in the parotid space include calculi, lymphoepithelial lesions associated with AIDS, lipomas, first branchial cleft cysts, and, rarely, schwannomas (Fig. 4-18) (266,339,369). Additionally, the parotid glands may painlessly enlarge in conditions such as chronic starvation, diabetes mellitus, and alcoholism, and with the use of certain drugs such as thiazide diuretics. Bilateral painful enlargement may be caused by viral illnesses such as mumps, whereas unilateral involvement is more commonly due to bacterial infection either with or without stone disease. Collagen vascular disease, such as Sjögren syndrome, and granulomatous disease, such as sarcoidosis and tuberculosis, may also enlarge the glands.

PARAPHARYNGEAL SPACE The parapharyngeal space is shaped like an inverted pyramid or tetrahedron and extends from the skull base to the hyoid bone (374,438) (see Fig. 4-7). This space is triangular on transaxial images with the apex pointing toward the nasopharynx. Anterolaterally, it is bounded by the medial pterygoid fascia, which separates it from the masticator space. Medially, the parapharyngeal space is bordered by the pharyngobasilar fascia. At the level of the nasopharynx,

B

A Figure 4-16 Basal cell monomorphic adenoma, deep parotid lobe. A well-circumscribed mass (white arrow) extends from the deep lobe of the left parotid gland into the prestyloid compartment of the left parapharyngeal space. The lesion is isointense to muscle on the T1-weighted image (A) and hyperintense on the fat-saturated, enhanced, T1-weighted image (B). A rim of parapharyngeal fat (arrowheads) surrounds the medial, anterior, and posterior margins of the tumor. The lateral border of the tumor abuts parotid tissue. At surgery, the lesion was found to arise from the deep lobe of the parotid. Parotid gland (P); retromandibular vein (black arrow).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 157

Neck

A

157

B Figure 4-17 Pleomorphic adenoma. A left parotid gland mass (arrows) occupies portions of both the superficial and the deep lobes. It is hypointense on the T1-weighted image (A) and hyperintense on the T2-weighted image (B). (From Brown J, Wippold FJ 2nd. Practical MRI: a teaching file. New York: Lippincott-Raven Press, 1996:200–220, with permission.)

this space is subdivided into prestyloid and poststyloid compartments by the fascia of the tensor veli palatini. At the level of the oropharynx, the oropharyngeal constrictors separate the compartments (70,343). The prestyloid compartment contains branches of the internal maxillary and ascending pharyngeal arteries, nerves, fat, salivary rests, and minor salivary glands (188). Primary lesions of the prestyloid compartment are often salivary in origin, such as pleomorphic adenomas of salivary rests (70). Benign salivary lesions tend to be well defined, nonenhancing or minimally enhancing, and isodense on CT images (361). Compared with muscle signal intensity,

these lesions are usually hypointense to isointense on T1-weighted MR images and isointense to hyperintense on T2-weighted MR images (368). Schwannomas may also arise within the prestyloid parapharyngeal space. These benign tumors are usually well marginated with crisply defined borders, hypointense on T1-weighted MR images, and markedly hyperintense on T2-weighted MR images. Other lesions within this compartment include lipomas, atypical branchial cleft cysts, metastases, and lesions spreading from adjacent compartments (Fig. 4-19) (18,94). Intrinsic parapharyngeal lesions have a surrounding halo of parapharyngeal fat, whereas deep

A

B Figure 4-18 Lymphoepithelial cysts in a patient with AIDS. Enhanced fat-saturation (A) and T2-weighted (B) magnetic resonance images demonstrate multiple cysts (arrowheads) within the parotid glands. The lesions are hypointense on T1-weighted images and hyperintense on T2-weighted images. Cervical lymphadenopathy (arrows) is also noted.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 158

158

Chapter 4

Figure 4-19 Lipoma, prestyloid parapharyngeal space. CT demonstrates a lipid density bilobed mass (asterisks) that conforms to the right prestyloid parapharyngeal space. The anterior border of the lesion outlines the medial pterygoid muscle (m), and the posterior border defines the poststyloid parapharyngeal space (carotid space). The mass extends laterally through the stylomandibular tunnel (dotted lines), which is defined as the space between the styloid process (arrowhead ) and the mandible (M), to deform the medial border of the right parotid gland (G). The right parotid gland is largely replaced by hypodense fat.

parotid lesions extending into the parapharyngeal space are inseparable from the parotid gland tissue (18,361,368). This differentiation is important in that surgeons prefer to approach primary parapharyngeal masses from the oral or submandibular regions and to use a parotid approach for deep parotid lobe lesions to conserve the facial nerve (368). Some asymmetry in the parapharyngeal space is acceptable; however, total asymmetric obliteration of parapharyngeal fat indicates a lesion (36,270). Parapharyngeal space masses tend to displace the carotid artery and jugular veins posteriorly (361,368). The poststyloid compartment is also known as the carotid space as it extends below the hyoid bone.

accessory nerve, hypoglossal nerve, sympathetic chain, and the internal jugular nodes of the deep cervical chain (70,111,343). Lesions occurring in this space arise from the tissues found within the space and secondarily from neighboring spaces. The carotid artery may be ectatic in older patients and deviate medially to present within the adjacent retropharyngeal region (Fig. 4-20) (336). Appreciation of this variant is essential in avoiding inappropriate biopsy. Imaging of the neck, especially with MR angiography or CT angiography, may reveal luminal stenoses and occlusions, pseudoaneurysms, and dissections (251). Spontaneous dissection of the internal carotid artery is a cause of cerebral ischemia in young patients and may be accompanied by unilateral headaches and Horner syndrome. The right jugular vein is frequently larger than its companion on the left and should not be mistaken for a mass (134). Paragangliomas are slowly growing, hypervascular lesions of neural crest origin (111). These tumors are classified by origin and location (421). Tumors involving the carotid bifurcation (carotid body tumors) are the most common type, followed by lesions involving nerves of the jugular bulb (glomus jugulare) and the nodose ganglion of the vagus nerve (glomus vagale) (Fig. 4-21). Symptoms depend on the location of the tumor and include pulsatile tinnitus and lower cranial neuropathy. Lesions are multiple in 5% of patients, and almost 30% of patients have a familial history of the disease (111). Paragangliomas often demonstrate permeative bone erosion on imaging. Glomus jugulare tumors located at the skull base usually destroy the jugular spine, a feature that differentiates them from schwannomas (172). Typically smoothly contoured, paragangliomas are usually of intermediate signal intensity on T1-weighted MR images compared with muscle and hyperintense on

CAROTID SPACE The cylindrical carotid space extends from the base of the skull to the aortic arch and is invested with all three layers of the deep cervical fascia (111,147,343) (see Fig. 4-7). In the suprahyoid neck, some authors refer to the carotid space as the poststyloid parapharyngeal space. The suprahyoid portion of the carotid space is bordered anteromedially by the pharynx, posteriorly by the perivertebral fascia, and anterolaterally by the prestyloid parapharyngeal space. In the infrahyoid region, this space is surrounded by the visceral and retropharyngeal spaces medially, the perivertebral and posterior cervical spaces posteriorly, and the sternocleidomastoid muscle anterolaterally (343). The carotid space contains the carotid artery, internal jugular vein, glossopharyngeal nerve, vagus nerve, spinal

Figure 4-20 Tortuous carotid arteries. Enhanced computed tomography demonstrates medial deviation of both internal common carotid arteries (arrows) into the retropharyngeal space. Internal jugular veins (arrowheads).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 159

Neck

159

B

A Figure 4-21 Carotid body tumor. A large, enhancing poststyloid parapharyngeal mass (arrows) is identified on the computed tomography scan (A). The internal architecture of the mass is irregularly hyperdense. The common carotid angiogram (B) reveals a hypervascular mass splaying the internal carotid (arrowheads) posteriorly and the branches of the external carotid (arrows) anteriorly.

T2-weighted MR images (364). Paragangliomas enhance vigorously on imaging and may show signal voids resulting from flow or tiny foci of hemorrhage on MR images (Fig. 4-22) (283,361,364,368,421). This “salt and pepper” appearance is more consistently seen with lesions greater than 2 cm in size (172). Residual mass and bone erosion may persist even after successful radiation treatment and do not necessarily imply tumor recurrence (260). A mimic of the glomus jugulare tumor is the extracranial meningioma arising in the jugular foramen (98,252). Compared with paragangliomas, meningiomas tend to enhance more homogeneously, are more likely to have an enhancing tail of dura at the skull base, and stain less vigorously on angiography (252). Nerve sheath tumors, such as schwannomas or neurofibromas, may involve the vagus nerve or the sympathetic plexus. Nerve sheath tumors displace the carotid artery anteriorly and medially (338,342). Consistent differentiation among the types of nerve sheath tumors on imaging is impossible. Schwannomas are usually well encapsulated and isodense to hypodense on CT imaging and hypovascular on angiography. Moreover, schwannomas tend to smoothly erode bone, in contradistinction to the permeative, irregular margins of destruction caused by paragangliomas (111,342). Hypervascular or malignant varieties are more difficult to differentiate from other vascular

tumors. Occasionally, neurofibromas may be hypodense or frankly cystic appearing on CT (368), although schwannomas also may undergo cystic or necrotic transformation. Either nerve sheath tumor may be associated with neurofibromatosis. Other lesions of the carotid space include lymphomas, branchial cleft cysts, and lipomas. Metastases and processes spreading from neighboring spaces also may involve this region. Because of the proximity of carotid space lesions to major vessels and because of the potential hypervascularity of some of the entities arising in this compartment, angiography is frequently used to preoperatively evaluate blood supply. Additionally, angiographically delivered embolization has emerged as an extremely useful tool for devascularizing lesions prior to surgery (176).

MASTICATOR SPACE The masticator space is formed by the splitting of the superficial layer of deep cervical fascia (see Fig. 4-7A). This suprahyoid space contains the mandible, the muscles of mastication, and the mandibular division of the trigeminal nerve (36,70,407,438). Lesions derived from these tissues include nerve sheath tumors, mandibular and soft tissue sarcomas, dental tumors, cysts and abscesses, osteomyelitis,

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 160

160

Chapter 4

Figure 4-22 Glomus jugulare tumor. Fat-saturated, T1-weighted magnetic resonance image following intravenous administration of gadopentetate dimeglumine reveals an intensely enhancing mass (arrowheads) in the right carotid space (also termed the poststyloid compartment of the parapharyngeal space). This lesion is homogeneous, although many glomus tumors demonstrate signal voids. The right internal jugular vein is filled with tumor, and the right internal carotid artery (white arrow) is displaced anteromedially. Styloid process (black arrow), parotid gland (P).

and lipomas (145,312,340) (Figs. 4-23 to 4-26). Vascular lesions, such as hemangiomas, lymphangiomas, and, rarely, pseudoaneurysms (see Fig. 4-1), may also arise in this space. In children, rhabdomyosarcoma also may present in the masticator space (46) (Fig. 4-27). Indeed, because of its prevalence in children and its tendency to arise in virtually any anatomic space, the diagnosis of rhabdomyosarcoma should be considered for any pediatric head and neck mass. Occasionally, tumors display a multiloculated ring enhancement pattern resembling bunches of grapes. This so-called botryoid sign may reflect a histologic feature in which mucoid-rich stroma is covered with tumor cells (144). Care must be taken to avoid misinterpreting this sign as evidence of infection. Rhabdomyosarcoma may also mimic a benign lesion by initially remodeling bone rather than frankly destroying it. Squamous cell carcinoma metastases from the oral mucosa and salivary glands also can spread into the masticator space (145). The mandibular branch of the trigeminal nerve exits the skull through the foramen ovale, which is located above the masticator space and has been termed the “chimney of the masticator space” (46). Lesions within the masticator space can invade the middle cranial fossa by this route, and intracranial processes, such as meningiomas, can

Figure 4-23 Facial cellulitis and dental abscess. Enhanced computed tomography image demonstrates extensive edema in the subcutaneous fat and soft tissues of the right neck. Minimal edema is also noted in the left face. This patient had a right molar extraction complicated by infection in the right masticator space that spread into adjacent compartments. The fluid collection (black arrowheads) adjacent to the cortex of the mandible is a loculated abscess. The right masseter (m) is swollen, and the enhancing parotid gland (g) is inflamed. A large enhancing internal jugular lymph node (asterisk) is evident.

descend into the masticator space and become extracranial (see Fig. 4-27). Signs of perineural spread along the mandibular division of the trigeminal nerve include: thickening of V3, expansion of the foramen ovale, mass within Meckel’s cave, lateral bulging of the cavernous sinus, and atrophy of the muscles of mastication (210).

PHARYNGEAL MUCOSAL SPACE The pharyngeal mucosal space is a term used to include the mucosal surfaces and immediate submucosa of the nasopharynx and oropharynx. The oral cavity and the suprahyoid portion of the hypopharynx can also be conveniently included in this discussion (see Fig. 4-7) (286). Most of this suprahyoid space is surrounded posteriorly and laterally by a sleeve comprised of the middle layer of the deep cervical fascia. Superiorly, this fascia envelopes the posterior aspect of the pharyngobasilar fascia, which attaches the pharynx and superior constrictor muscle to the base of the skull (211,228,286). Also included in this space are lymphoid tissue, minor salivary glands, pha-

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 161

161

Neck

B

A Figure 4-24 Trigeminal nerve schwannoma. Enhanced T1-weighted (A) and T2-weighted (B) images reveal a large dumbbell-shaped mass within the right masticator space. The posterior wall of the right maxillary sinus (arrows) and the zygomatic arch (arrowheads) are remodeled by this slowly growing lesion. The central component is hypointense on both the T1-weighted and T2-weighted sequences. The peripheral portion of the lesion enhances and is hyperintense on the T2-weighted image. This appearance has been described as a target sign and may be seen in nerve sheath tumors. (From Brown J, Wippold FJ 2nd. Practical MRI: a teaching file. New York: Lippincott-Raven Press, 1996:200–220, with permission.)

ryngeal constrictor muscles, and the salpingopharyngeus muscle (400). The nasopharyngeal portion of the pharyngeal mucosal space extends from the posterior boundary of the nasal cavity to a plane defined by the hard and soft palate. The nasopharynx includes the vault, lateral walls, and posterior wall (137). The oropharyngeal portion extends from the inferior margin of the nasopharynx to the level of the pharyngoglottic and glossoepiglottic folds. It consists of the pharyngeal wall between the nasopharynx and the pharyngoepiglottic fold, the soft palate, the tonsillar region, and the tongue base (137,254,255). The oral cavity consists of the floor of the mouth, the anterior two thirds of the tongue, the buccal mucosa and gingiva, the hard palate, and the retromolar trigone. The hypopharynx is considered in the section on the visceral space because of its relationship with the larynx. Predictably, the most common malignant lesion of the pharyngeal mucosal space is carcinoma (Fig. 4-28). Risk factors include exposure to tobacco and alcohol. Genetic factors and possibly clinical carcinogens also are suspected (64,201,426). In the oral cavity, syphilis is an additional risk (245). Carcinoma of the nasopharyngeal mucosa differs from cancers elsewhere in the pharyngeal mucosal space in that it favors patients of Chinese ancestry, has less association with the risk factor of smoking, and may be associated with the Epstein-Barr virus (22,64,211,426,453). Nasopharyngeal cancer may be keratinizing, nonkeratinizing, or undifferentiated (426). These tumors tend to burrow

into the submucosal tissues and may threaten the skull base prior to significant symptom development. On imaging, small tumors may remain subtle. Obliteration of the fossa of Rosenmüller may provide an early clue to tumor presence. Serous otitis media due to eustachian tube dysfunction should also alert the radiologist to the lesion. A radiosensitive subgroup of nasopharyngeal cancer is known as lymphoepithelioma (Schminke tumor) (211). Carcinomas of the pharyngeal mucosal space may erode branches of the external carotid artery and cause cataclysmal hemorrhage. CT and/or MRI are rarely useful in this situation except to verify a previously unsuspected tumor. Angiography is the preferred procedure. Intraarterial embolization is an effective method of controlling the bleeding (440). The lymphoid tissue of the pharyngeal mucosal space includes the adenoids and tonsils and is collectively named Waldeyer’s ring. Normally, the lymphoid tissue is asymmetric and involutes with age; therefore, the consideration of neoplasm in a young patient must be carefully weighed with a history of recent upper respiratory infection and the likely presence of normal variability (245, 270,425). Non-Hodgkin lymphomas develop from this tissue (211). Inflammatory lesions such as tonsillar abscess also are fairly common and can present with sore throat, high fever, and a mass in the tonsillar region. Postinflammatory calcifications frequently are seen as incidental findings on CT (Fig. 4-29).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 162

162

Chapter 4

A

B Figure 4-25 Osteosarcoma, masticator space. T1-weighted transaxial magnetic resonance (MR) image (A) demonstrates an enhancing hetergeneous mass (arrows) that straddles the left mandible (arrowheads) and splays the muscles of mastication (M). The lesion (arrows) is hyperintense on the T2-weighted coronal MR image (B).

Minor salivary tumors also are seen in this region, especially involving the mucosal surfaces of the hard and soft palates. These lesions may be benign, such as pleomorphic adenomas, or malignant, such as adenocystic carcinomas. Pleomorphic adenomas are often well circumscribed and appear as submucosal lesions. Occasionally, a minor salivary gland becomes obstructed and produces a small retention

cyst (67). In children, rhabdomyosarcomas may invade the nasopharyngeal region and infiltrate the mucosa. In young male patients, angiofibromas also may involve the nasopharyngeal region. Mucosal space lesions usually are well seen by the clinician; however, imaging evaluates deep invasion, bone destruction, orbit and sinus involvement, intracranial extension,

B

A Figure 4-26 Hemangioma. A large mass (white arrowheads) occupies the right masticator space. The lesion is heterogeneous on the computed tomography image (A) and heterogeneously hyperintense on the T2-weighted magnetic resonance image (B). The lesion straddles the mandible (arrows) and displaces the masseter muscle laterally (black arrowheads).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 163

163

Neck

Figure 4-29 Tonsillar calcifications. Calcifications (arrowheads) identified in the tonsils on computed tomography.

Figure 4-27 Rhabdomyosarcoma. Enhanced T1-weighted coronal magnetic resonance image demonstrates a large, enhancing soft tissue mass (arrowheads) arising in the right masticator space and enveloping the mandible. The lesion extends through the expanded foramen ovale (arrow) into Meckel’s cave. (From Brown J, Wippold FJ 2nd. Practical MRI: a teaching file. New York: LippincottRaven Press, 1996:200–220, with permission.).

and lymph node metastases (320,332,333). MRI is an excellent modality because of its superb display of tissue contrast, its ability to portray coronal and transaxial planes, its independence from dental amalgam artifacts, and its ability to depict spread into adjacent anatomic regions (82,390,418). Moreover, imaging results in upstaging almost 25% of advanced pharyngeal mucosal space cancers

by demonstrating clinically silent submucosal disease (120). On CT or MR images, the mucosa may be thickened and asymmetric. As previously mentioned, nasopharyngeal cancers typically blunt the fossa of Rosenmüller and displace the parapharyngeal fat (Figs. 4-30 and 4-31) (64,167). The retropharyngeal and perivertebral spaces may be thickened, and the sphenoid and ethmoid sinuses may be invaded (333). Extension along the eustachian tube, the pterygoid canal, and the parapharyngeal fat planes jeopardizes the skull base and intracranial cavity (211,401). Associated serous otitis media resulting from eustachian tube obstruction is detected on imaging as

A

B Figure 4-28 Squamous cell carcinoma of the tongue. A right tongue mass (arrows) is revealed on transaxial magnetic resonance images. The lesion is mildly hyperintense on the enhanced T1weighted image (A) and hyperintense on the T2-weighted image (B). The lesion displaces the midline septum of the tongue to the left (arrowheads). Anatomic asymmetry of the tongue and oral cavity is an important clue to pathology.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 164

164

Chapter 4

Figure 4-30 Nasopharyngeal carcinoma. T1-weighted magnetic resonance image demonstrates a large nasopharyngeal mass (arrows). The lesion invades the retropharyngeal and perivertebral spaces posteriorly and involves the anterior arch of the C-1 vertebra. The mass obliterates the fat of the left parapharyngeal space and displaces the left masticator space contents anterolaterally. The mass invades the masticator space and abuts the left mandibular condyle (black arrowhead). Anteriorly, the lesion involves the pterygoid process, left pterygopalatine fossa, and posterior nasal cavity. Note the soft tissue (white arrowhead) in the left mastoid tip resulting from eustachian tube obstruction.

caused by a Pseudomonas infection in a diabetic patient. Oral cavity and oropharyngeal cancers may spread to involve the mandible. Rarely, lymphoma may primarily involve the mandible. The presence of cortical breakdown and abnormal marrow signal is a highly sensitive MR indicator of periosteal/cortical and medullary invasion (Fig. 4-32). Unfortunately, radiation changes in the marrow, osteoradionecrosis, and osteomyelitis may produce similar findings (66). The tongue, which straddles both the oral cavity and oropharynx, is a frequent site of pathology. Most patients referred for imaging have biopsy-proved carcinoma, and MR studies are needed to establish the extent of deep invasion. Benign lesions do occur, however, and hemangiomas, dermoids, epidermoids, lingual thyroid tissue, and thyroglossal duct cysts should also be considered, especially for submucosal lesions. An often overlooked finding is that of tongue atrophy, which may signal hypoglossal nerve trauma or perineural cancer invasion or even an unsuspected skull base lesion. Following nerve injury, the tongue loses bulk and the muscle is replaced by fat.

VISCERAL SPACE fluid-filled middle ear and mastoid air cells (344). MRI is especially useful in establishing soft tissue and dural invasion, although CT is useful in depicting bone invasion at the base of the skull (4,282,333). Careful clinical correlation is also essential, because non-neoplastic lesions, such as malignant otitis externa (MOE), can mimic the appearance of nasopharyngeal cancer. MOE is typically

The midline visceral space is enclosed by the middle layer of deep cervical fascia and extends from the hyoid bone to the mediastinum (see Fig. 4-7C). It contains the larynx and hypopharynx, the thyroid and parathyroid glands, the trachea and esophagus, paratracheal lymph nodes and the recurrent laryngeal nerves (36).

A

B Figure 4-31 Nasopharyngeal carcinoma. Large nasopharyngeal mass is hypointense on transaxial T1-weighted (A) and hyperintense on enhanced, sagittal T1-weighted (B) MR images. The lesion displaces the internal carotid arteries (arrows) and jugular veins (arrowheads) laterally. The mass invades the perivertebral space and the clivus (open arrows).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 165

Neck

Figure 4-32 Lymphoma of the mandible. T1-weighted magnetic resonance image reveals diffuse soft tissue mass (black arrows) replacing the normal hyperintense marrow fat signal (asterisk) and extending into the extraosseus fat anterior to the mandible. Portions of the cortex are destroyed (white arrows).

Larynx and Hypopharynx The larynx and hypopharynx dominate the visceral space. As the sentry of the respiratory tract, the larynx prevents aspiration during swallowing. The larynx also permits phonation. Closure of the glottis permits efficient coughing and removal of secretions and foreign matter. Glottis closure also facilitates elevation of the intra-abdominal pressure, aiding expulsion of rectal and uterine contents. The hypopharynx is the portal to the digestive tract and channels food into the esophagus. Both CT and MRI demonstrate the larynx and hypopharynx well (Figs. 4-33 to 4-36) (15,53,221,223,376). CT better depicts calcified cartilage compared with MRI, but MRI better defines soft tissue planes. Swallowing and vascular pulsations may degrade MR images.

Anatomy of the Larynx and Hypopharynx The Hyoid Bone and Laryngeal Cartilages The laryngeal skeleton provides the framework for the larynx and includes the hyoid bone and the laryngeal cartilages (460) (Figs. 4-37 and 4-38). The hyoid bone is a U-shaped structure that is anatomically considered part of the lingual apparatus, but provides a scaffold for muscles that suspend the larynx. The hyoid bone consists of a central body and paired lateral greater and lesser horns. Embryologically, the lesser horns are derived from the second branchial arch and the body and greater horns are derived from the third branchial arch. The second branchial arch also forms the styloid process, stylohyoid ligament, and thestapes bone. Congenital anomalies of the hyoid bone should prompt inspection of the middle ear and vice versa.

165

The line of fusion of the body and greater horns of the hyoid bone should not be mistaken for a fracture (see Fig. 4-33). Positioned caudal to the hyoid bone, the thyroid cartilage is the largest laryngeal cartilage. The two ala of the thyroid cartilage fuse anteriorly to form a V-shaped shield. The anterior vertical fusion line forms the ridge of the laryngeal prominence. The naturally occurring notch in the superior portion of the laryngeal prominence should not be misinterpreted as cancer-related cartilage destruction on transaxial imaging studies. The superior and inferior cornua project from the posterior free edges of the thyroid cartilage. The superior cornua attaches to the hyoid bone by the thyrohyoid ligament, and the inferior cornua articulates to the cricoid cartilage by a synovial joint (34,165). The imaging appearance of the thyroid cartilage varies considerably because of its hyaline composition and potential for mineralization (12,231,449) (Fig. 4-39). Calcification occurs by 8 to 10 years of age and most commonly begins with the posterior portions of the cartilage (406). Extensive calcification before 13 years of age is unusual, but may be seen with cardiovascular disease, skeletal dysplasias, warfarin therapy, and hypercalcemic states. Normal mineralization may be quite irregular, is often asymmetric, generally progresses at a variable rate throughout life, and is especially pronounced in older men (11). A fat-containing medullary space may form during calcification and ossification. The irregular mineralization should not be confused with tumor invasion. The cricoid cartilage is the only cartilage that forms a complete ring and is therefore the foundation and linchpin of the entire larynx. It resembles a signet ring with the broad lamina located posteriorly and the narrow rim sloping anteriorly. Calcification of the cricoid cartilage is usually more symmetric than that of the thyroid cartilage. The cricoid cartilage nestles beneath the thyroid cartilage and is suspended from the thyroid cartilage by the cricothyroid membrane. The cricoid cartilage attaches to the inferior cornua of the thyroid cartilage by synovial joints that effect an approximate 1.5-mm separation between the two cartilages. These symmetric cricothyroid joints hinge both cartilages and permit the anterior cricoid arch to pivot superiorly toward the thyroid lamina (34,165). As with other synovial joints, the cricothyroid joints are susceptible to arthritis (43). Patients with cricothyroid arthritis may present with odynophonia and demonstrate vocal cord bowing and inability to achieve vocally high pitch on physical examination (136,229). The paired pyramid-shaped arytenoid cartilages perch atop the posterosuperior rim of the cricoid cartilage and are affixed by synovial joints. The base of each arytenoid cartilage has an anterior vocal process composed of elastic cartilage. The posterior portion of the true vocal cords attaches to these vocal processes. Posterolateral muscular processes of hyaline cartilage serve as sites of attachment for the lateral cricoarytenoid, posterior cricoarytenoid,

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 166

166

Chapter 4

A,B

C

D,E

F

I

G,H Figure 4-33 Normal computed tomography (CT) anatomy of the larynx: transaxial CT slices arranged cephalocaudad. A: The suprahyoid portion of the epiglottis (white arrow) is seen posterior to the valleculae and anterior to the laryngeal vestibule; internal carotid artery (C); internal jugular vein (J); submandibular gland (G); anterior belly of digastric muscle (A); mylohyoid and geniohyoid muscles (M); sternocleidomastoid muscle (SCM); posterior pharyngeal wall (black arrows); normal high internal jugular lymph node (arrowhead). B: The body (Hb) and a portion of the right greater cornu (black arrow) of the hyoid bone are visible. The two air-filled valleculae (V ), which can normally be asymmetric, are separated by the median glossoepiglottic fold (white arrow). The low-attenuation zone posteriorly within the right carotid artery represents atherosclerotic plaque (black arrowhead); epiglottis (white arrowhead); vestibule (Ve), mylohyoid and geniohyoid muscles (suprahyoid strap muscles) (M). C: The valleculae (v) are smaller in size. Normal lucency (white arrowhead) between the body and greater cornus of the hyoid bone is seen; omohyoid muscle (large arrow); platysma muscle (small arrows). D: The fat-containing preepiglottic space (PES) is seen anterior to the soft tissue–density epiglottis (small white arrowhead) and extends laterally into the paralaryngeal space (PLS) and then posteriorly into the aryepiglottic folds (arrows); pyriform sinus (P); superior cornu of thyroid cartilage (black arrowhead ); normal spinal accessory lymph node (large white arrowhead); preepiglottic space (PES). E: The left aryepiglottic fold (arrow) is more clearly seen separating the laryngeal vestibule (Ve) and left pyriform sinus (P). The aryepiglottic folds may normally be asymmetric in thickness. F: Just above the thyroid notch, the epiglottis (large arrow) has tapered and is not nearly as broad as its suprahyoid portion. A small amount of air (arrowhead) is seen within a minimally dilated left saccule of the laryngeal ventricle; infrahyoid strap muscles (S). G: The false vocal cord (small arrows), which lies medially to the fat-containing paralaryngeal space, is seen at the level of the base (arrowhead) of the arytenoid cartilage and the thyroid notch (large arrow). Incomplete mineralization of the thyroid lamina is a normal variation and should not be mistaken for cartilage

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 167

Neck

167

The cuneiform cartilages form bulges in the posterolateral rims of the aryepiglottic folds and are a source of support (Fig. 4-42). Finally, the triticeus cartilages form bulges in the posterior free margin of the thyrohyoid membrane (136,406).

Laryngeal Ligaments and Membranes J Figure 4-33 (continued ) destruction: common carotid artery (C); internal jugular vein (J); infrahyoid strap muscle (S); sternocleidomastoid muscle (SCM). H: The anteromedially projecting vocal process (arrowhead ) of the arytenoid cartilage (A), which articulates with the cricoid cartilage (Cr), demarcates the level of the true vocal cord (arrows). The thyroid lamina have fused anteriorly to form the laryngeal prominence (P). The soft tissues at the anterior commissure just posterior to the prominence should normally be less than 1 mm in thickness. The laryngeal airway is elliptical in shape, with a long anteroposterior axis, and also closely abuts the cricoid cartilage posteriorly. I: At the subglottic level, the cricoid cartilage (black arrowheads), consisting of a dense cortical rim with a lower-density medullary space, is seen. The cricothyroid joints (arrows) are symmetric bilaterally. The undersurfaces of the true vocal cords (white arrowheads) are seen anterolaterally; common carotid artery (C); internal jugular vein (J). J: On the next caudal image, the cricoid ring is about two-thirds complete, except anteriorly at the cricothyroid membrane (white arrowhead ). The subglottic airway is more circular than at the true-vocal-cord level and is closely apposed to the cricoid cartilage; inferior cornua of thyroid cartilage (arrows); cricopharyngeus-cervical esophagus (black arrowheads); thyroid gland (Th).

transverse arytenoid, and oblique arytenoid muscles (34,165). The lateral muscular processes and body of the arytenoid cartilage calcify with age (136). The muscular processes should be no more than 2 mm from the inner surface of the thyroid lamina. The epiglottis is a leaflike cartilage that hovers over the glottic inlet forming the anterior wall of the laryngeal vestibule. The epiglottis consists primarily of elastic cartilage and seldom calcifies (Fig. 4-40). This cartilage is draped anteriorly by stratified squamous epithelium and posteroinferiorly by pseudostratified, ciliated, and columnar epithelium. Inferiorly, the apex of the base forms a narrow stem or petiole that attaches to the posterior surface of the angle of the thyroid cartilage immediately above the anterior attachment of the true vocal cords. Superiorly, the anterior surface of the epiglottis attaches to the posterior surface of the body of the hyoid bone by the hyoepiglottic ligament (34,165) (Fig. 4-41). The transaxial plane of the hyoid bone divides the epiglottis into the suprahyoid and infrahyoid regions. Recognizable mucosal folds that drape the anterior surface include the median glossoepiglottic fold and the paired lateral glossoepiglottic folds. These folds form the valleculae (136,406). The tiny corniculate cartilages crown the apices of the arytenoid cartilages and articulate by synovial joints. The aryepiglottic folds attach to the corniculate cartilages.

Binding the cartilages is a system of ligaments and membranes that provide much of the structural and functional integrity of the larynx (460). These soft tissues form the true vocal cords, false vocal cords, and aryepiglottic folds. One of the most important of these membranes is the conus elasticus, also known as the triangular membrane (see Fig. 4-38). Although not usually identified on CT, the conus elasticus contributes to the structure of the true vocal cords. The conus elasticus resembles an inverted funnel. The thickened superior edges form the vocal ligaments. The vocal ligaments together with the thyroarytenoid and vocalis muscles form the true vocal cords (136,224,307). Controlled vibration of the true vocal cords is responsible for speech. The vocal ligaments attach anteriorly to the inner surface of the thyroid cartilage prominence, forming the anterior commissure. The soft tissues posterior to the thyroid cartilage at the anterior commissure should measure no more than 1 mm in thickness. Unusual fullness of tissue in this area may signify cancer of the anterior true cords and contralateral spread. Careful attention to imaging technique is mandatory in order to avoid false–positive interpretations. Improper neck positioning and scanning angulation, which may inadvertently project the normally thicker anterior false cords into the same tissue plane as the posterior true cords, and adduction of the true cords may contribute to erroneous diagnoses. The posterior vocal ligaments attach to the vocal processes of the arytenoid cartilages (323). The conus elasticus membrane sweeps inferolaterally toward the cricoid cartilage. The base of the conus elasticus then splits so that a superior slip attaches to the superior rim of the cricoid cartilage and an inferior slip attaches to the inferior margin of the cricoid cartilage (136,224,229). In addition to providing a foundation for the vocal cords, the tough conus elasticus influences the route of spread of carcinomas of the larynx (179). The paired quadrangular membranes attach to the lateral free edges of the epiglottis and to the arytenoid cartilages, thus forming two curtains of tissue between the epiglottis and the arytenoid cartilages. The superior free margins run obliquely from the epiglottis to the arytenoids and are thickened to form the aryepiglottic folds. The medial surface of the folds is considered endolaryngeal and the lateral surface hypopharyngeal (324). The thickened inferior free margin stretches from the base of the epiglottis to the superior portion of the arytenoids and forms the false vocal cords (136,229). Separating the false and true vocal cords is a transversely oriented diverticulum

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 168

168

Chapter 4

C

A,B

D,E

F

I

G,H Figure 4-34 Normal magnetic resonance (MR) anatomy of the larynx: transaxial T1-weighted MR images arranged cephalocaudad. A: The medium-signal-intensity epiglottis (arrow) seen posterior to the air-containing valleculae (V) is slightly higher in signal intensity than the sternocleidomastoid muscle (SCM ); glossoepiglottic fold (arrowhead). B: On a more caudal image in another patient, the hyperintense pre-epiglottic space (PES) is well seen adjacent to the medium-signal-intensity infrahyoid strap muscles (S). The aryepiglottic folds (arrowheads) are slightly more hyperintense than the epiglottis (arrow) because of their fat content; common carotid artery (C); internal jugular vein (J); pyriform sinus (P). C: At the level of the false vocal cords, the paralaryngeal space and false vocal cords (small arrowheads) are relatively hyperintense because of loose areolar and glandular tissue. The calcified portion of the thyroid lamina is seen as a hypointense zone (arrows) adjacent to the hyperintense medullary fat; thyroid notch (large arrowhead ). D: In another patient at the same level, a more substantially calcified thyroid cartilage appears as a black strip separating the medium-signalintensity infrahyoid strap muscles (arrows) and hyperintense paralaryngeal space and false vocal cord (arrowheads). E: Hyperintense medullary fat within portions of the thyroid lamina (arrows) is seen in a different patient. Note how other areas of the thyroid cartilage blend in with the infrahyoid strap muscles (S). F: At the level of the true vocal cords, the signal intensity of the true cords (arrowheads) is slightly higher than that of the strap muscles. a, vocal process of arytenoid cartilage (a); top of cricoid cartilage (Cr); common carotid artery (C); internal jugular vein (J); thyroid lamina (arrows). G: In another patient, hyperintense fatty marrow within the thyroid cartilage (large arrows), arytenoid cartilage, and cricoid cartilage (small arrows) sharply contrasts with the hypointense strap muscles and true vocal cords. H: At the subglottic level, the cricoid cartilage (arrows) is well visualized secondary to fat within its medullary cavity; undersurfaces of the vocal cords (small arrowheads); common carotid artery (C); internal jugular vein (J). I: On image 1 cm caudal, the airway is closely apposed to the cricoid ring. The thyroid gland (t) is intermediate in signal intensity between muscle and fat; esophagus (e). (C From Glazer HS, Niemeyer JH, Balfe DM, et al. Neck neoplasms: MR imaging. Part I. Initial evaluation. Radiology 1986;160:343–348, with permission.)

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 169

Neck

A

169

B

Figure 4-35 A: Coronal T1-weighted magnetic resonance images of the larynx. B: successive dorsal images; cricoid cartilage (large black arrow); thyroid cartilage (small black arrow); laryngeal ventricle (small black arrowheads); arytenoid cartilage (small white arrowhead ); aryepiglottic fold (small white arrow); pyriform sinus (P), paralaryngeal space (PLS); vestibule (V); strap muscles (S).

known as the laryngeal ventricle (Fig. 4-43). The lateral and upward extension of the ventricle is known as the saccule. Coronal MRI demonstrates the ventricle and saccule better than transaxial CT. Other ligaments and membranes include the thyrohyoid membrane, which attaches the hyoid bone to the thyroid cartilage, and the cricothyroid membrane, which connects the cricoid cartilage to the thyroid cartilage. Fibers from this latter membrane contribute to the anterior portion of the conus elasticus.

Laryngeal Muscles The muscles of the larynx orchestrate some of the most complex movements found anywhere in the body (460). These muscles are generally named for their cartilages of origin and insertion, thereby making recollection of the anatomy less daunting. The muscles are divided into extrinsic and intrinsic groups. The extrinsic muscles are found external to the

Figure 4-36 Sagittal T1-weighted magnetic resonance image of the larynx; epiglottis (white arrow); cricoid cartilage (white arrowhead); vallecula (black arrowhead ); hyoid bone (H); mylohyoid and geniohyoid muscles (M); preepiglottic space (PES).

larynx proper, and in turn, are divided into suprahyoid and infrahyoid groups. The suprahyoid muscles originate from the base of the skull and insert on the hyoid bone. These muscles suspend the larynx and include the stylohyoid, both bellies of the digastric, and the geniohyoid. Contraction elevates the hyoid bone and larynx. The infrahyoid muscles originate primarily from the sternum and other infrahyoid structures and insert on the hyoid bone and thyroid cartilage. Contraction of the omohyoid, sternohyoid, and sternothyroid muscles depresses the hyoid bone and larynx. Contraction of the thyrohyoid muscle draws the hyoid bone and thyroid cartilage together (Fig. 4-44) (34,165,326). Intrinsic muscles are found within the larynx proper and can be subdivided into the vocal cord abductors/ adductors, tensors, and protectors (Fig. 4-45). The paired posterior cricoarytenoid muscles arise from the posterior cricoid cartilage and insert on the muscular processes of arytenoid cartilages. During contraction, the muscular processes pivot and open the glottis. These muscles are the only abductors of the vocal folds (34,165,326). The adductor group closes the glottic aperture and consists of the lateral cricoarytenoid, transverse arytenoid, oblique arytenoid, and thyroarytenoid muscles. The paired lateral cricoarytenoid muscles originate on the cricoid cartilage and insert on the muscular processes of the arytenoids. Contraction pivots the arytenoids such that the vocal cords close. Action of these muscles opposes the posterior cricoarytenoid muscles. Both the transverse and paired oblique arytenoid muscles join the two arytenoids and pull the cartilages together during contraction. The thyroarytenoid muscles join the arytenoid cartilages to the thyroid cartilage. The medial fibers insert on the vocal processes of the arytenoids and are termed the vocalis muscles (34,165,326). The paired cricothyroid muscles arise laterally from the thyroid cartilage and insert on the cricoid cartilage. These muscles lengthen, stretch, and tense the vocal cords by tilting the anterior cricoid cartilage upward and the posterior cricoid downward. Decreased tension on the vocal cords is

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 170

170

Chapter 4

A

B

C

Figure 4-37 Schematic drawings of the laryngeal skeleton. A: Frontal view. B: Lateral view. C: Lateral view after removal of right thyroid lamina.

A

B Figure 4-38 Section through the laryngeal skeleton and surrounding soft tissues. A: Coronal view. B: Sagittal view. Conus elasticus (CE).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 171

Neck

171

A,B

C Figure 4-39 Variation in thyroid (large arrows) and cricoid (small arrows) cartilage mineralization in an 18-year-old man (A), a 21-year-old woman (B), and a 50-year-old man (C ) . Paramedian thinning (arrowheads) normally can be seen.

accomplished by the thyroarytenoid muscle (34,165,326). Protection of the laryngeal inlet is accomplished by the aryepiglottic muscles, suspended between the arytenoid cartilages and the aryepiglottic fold, and the thyroepiglottic muscle, which travels from the thyroid cartilage to the epiglottis (34,165,326). All of the intrinsic muscles, with the exception of the cricothyroid muscle, are innervated by the recurrent laryngeal nerve, a branch of the vagus nerve. The cricothyroid muscle is innervated by the external branch of the superior laryngeal nerve, also a branch of the vagus. Sensation above the vocal cords is supplied by the internal laryngeal branch of the superior laryngeal nerve. Below the vocal cords, the recurrent laryngeal nerve provides afferent fibers (3). The right recurrent laryngeal nerve passes around the right subclavian artery and ascends in the right tracheoesophageal groove. The left recurrent laryngeal nerve loops under the aortic arch and then ascends in the left tracheoesophageal groove. The recurrent nerves enter the larynx deep to the inferior pharyngeal constrictor. These nerves are in close proximity to the thyroid gland.

Spaces within the Larynx Padding the cartilages, membranes, and muscles are several fibrofatty spaces that lie concealed from the view of the laryngoscopist, but are important in the spread of cancer (460).

The paraglottic (paralaryngeal) spaces are paired lateral spaces bordered medially by the quadrangular membrane, laterally by the thyrohyoid membrane and thyroid cartilage, posteriorly by the pyriform sinuses, and inferiorly by the conus elasticus. The laryngeal ventricles lie buried within this space. The paraglottic spaces are primarily fat filled and therefore are easily recognized on CT and MR images. The paraglottic spaces are wider at the level of the false cords and then narrow inferiorly at the level of the true cords (126). The midline anterior pre-epiglottic space is bordered superiorly by the valleculae, anterosuperiorly by the base of the tongue, anteriorly by the hyoid bone, inferiorly by the epiglottis, and laterally by the paraglottic space, thyrohyoid membrane, and the thyroid cartilage. The preepiglottic space extends from the hyoid bone to the anterior commissure of the true cords. Within this space are the hyoepiglottic ligament and glands. The pre-epiglottic space is primarily fat filled and especially well depicted on sagittal T1-weighted MR images.

Mucosal Surfaces of the Larynx The laryngeal mucosa drapes over the ligaments and cartilages to form a thin layer of folds. The midline glossoepiglottic fold covers the glossoepiglottic ligament, and the paired lateral

Figure 4-41 Hyoepiglottic ligament (large arrow) is seen within Figure 4-40 Calcification in epiglottic cartilage, an unusual finding in this elastic cartilage (arrow).

the preepiglottic space; infrahyoid strap muscles (S); epiglottis (arrowhead); aryepiglottic folds (small arrows). This midline structure should not be interpreted as tumor infiltration within the preepiglottic space.

5063_Lee_Ch04pp0145-0224 10/13/05 4:50 PM Page 172

172

Chapter 4

Figure 4-42 Cuneiform cartilage (white arrowhead); faintly visible corniculate cartilage (black arrowhead ) just cephalad to arytenoid cartilage; aryepiglottic fold (arrow); pyriform sinus (P).

Figure 4-44 Diagram of the extrinsic muscles of the larynx. The

pharyngoepiglottic folds extend from the pharyngeal wall to the epiglottis. These folds form the paired valleculae, which are cuplike spaces anterior to the epiglottis. The valleculae are often asymmetric and sometimes are intermittently coated with secretions, making detection of small cancers difficult (270).

Hypopharynx The hypopharynx is the portion of the foregut that lies between the oropharynx and the larynx, extending from the hyoid bone to the cricopharyngeus muscle. It is bounded anteriorly by the paraglottic space and laterally by the thyrohyoid membrane and the thyroid lamina. The hypopharynx consists of three regions: the paired pyriform sinuses, the postcricoid region (pharyngoesophageal junction), and the posterior hypopharyngeal wall. The pyriform sinuses are shaped like inverted cones with the apices directed inferiorly. The apex of each pyriform sinus lies lateral to the cricoarytenoid joint and above the cricothyroid space. The aryepiglottic folds separate the laryngeal vestibule from the pyriform sinuses. The pyriform sinuses end in the postcricoid region where the hypopharynx empties into the cervical esophagus (2,238,324). The postcricoid region extends from the level of the arytenoid cartilages to the inferior border of the cricoid cartilage

Figure 4-43 Laryngeal ventricle (arrows); arytenoid cartilages (A); cricoid cartilage (CR).

suprahyoid muscles, including the anterior and posterior bellies of the digastric muscle and the geniohyoid and stylohyoid muscles, lift the hyoid bone and larynx. The infrahyoid muscles, including the sternothyroid, sternohyoid, and omohyoid muscles, depress the larynx. The thyrohyoid muscle shortens the distance between the hyoid bone and the thyroid cartilage.

and is notoriously difficult to adequately assess by indirect or direct endoscopy (2). The posterior pharyngeal wall descends from the level of the hyoid bone to the inferior border of the cricoid cartilage. During swallowing, food channels laterally into the pyriform region, then posteriorly to the cricoid cartilage, and finally into the cervical esophagus, thus avoiding aspiration into the upper airway (2,238,324).

Airway At the level of the epiglottis and aryepiglottic folds, the airway is elliptical in shape. Descending toward the false cords, the airway narrows and assumes a tear-drop shape. The airway is elliptical at the true cords. The rima glottidis is the term used for the airway at the level of the true vocal

Figure 4-45 Diagram of the intrinsic muscles of the larynx as viewed from above, anterior at top of illustration. The triangular arytenoid cartilages (arrowheads) perch atop the cricoid cartilage. The ring-shaped cricoid cartilage is nestled within the shield-shaped thyroid cartilage. Thyroarytenoid muscles form the bulk of the true vocal cords. Vocal cord adductors include the oblique and transverse arytenoid muscles, lateral cricoarytenoid muscles, and the thyroarytenoid muscles. The posterior cricoarytenoid muscles abduct the cords by pivoting the arytenoid cartilages. The cricothyroid muscles lift the anterior ring of the cricoid cartilage and lengthen the vocal cords.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 173

Neck

cords. The intermembranous portion of the rima glottidis (glottis vocalis) consists of the ventral 60% of the cords, and the dorsal intercartilaginous portion (glottis respiratoria) consists of the portion between the arytenoid cartilages (225). Below the cricoid cartilage, the airway assumes a circular appearance. The posterior membrane of the trachea may posteriorly flatten, and the normal esophagus occasionally indents the airway silhouette (114). Nodular projections into the airway or asymmetric tracheal wall thickening should be viewed suspiciously as signs of subglottic cancer.

Pathology of the Larynx and Hypopharynx Carcinoma Overwhelmingly, most referrals for laryngeal and hypopharyngeal imaging concern the evaluation of masses. Masses within this region are usually malignant, and virtually all malignancies are squamous cell carcinomas arising from the aerodigestive mucosa (50). Lesions occur in men almost five times more commonly than in women and are almost always associated with use of tobacco and alcohol (325,426). The peak incidence is in the seventh decade of life (325). Rare cases of childhood laryngeal carcinoma have been reported; these are often associated with a predisposing factor of irradiation of a benign throat lesion, such as laryngeal papillomatosis (244). For staging purposes, the larynx is anatomically divided into the supraglottic, glottic, and subglottic (infraglottic) regions based on embryologic development and lymphatic drainage (24,69,387,408). Because of the proximity of the hypopharynx to the larynx, hypopharyngeal carcinomas are usually included in discussions of tumors in this region. As in other regions of the head and neck, staging is largely based on tumor size and involvement of anatomic subareas or regions. The TNM staging system requires determining the extent of the primary tumor (T), an assessment of lymphadenopathy (N), and determining whether or not distant metastases are present (M). In general, T1 tumors are small lesions confined to one anatomic subsite. T2 and T3 lesions are bulkier and often have spread to other subsites. T4 stage usually implies involvement of an adjacent structure (137,224,426). The clinical endoscopist usually makes the diagnosis of laryngeal carcinoma prior to referral for imaging (291). Endoscopy has several limitations, however, such as its inability to assess cartilage invasion, underestimation of submucosal spread, and inability to fully visualize lesions arising in the laryngeal ventricles and subglottic regions (383). Moreover, the clinical examination may fail to detect lymphadenopathy. Therefore, the role of the radiologist is to describe any deep extension, the relation of the mass to surrounding structures, and lymphadenopathy. Knowledge of patterns of spread of disease assists in appropriate evaluation of imaging findings. In general, imaging is usually not indicated in lesions clinically staged at T1 (50), although imaging is useful in

173

differentiating stage T2 from higher stages (291). CT has been well validated in laryngeal cancer staging (13,14,225, 329) and now has replaced laryngography, which, like laryngoscopy, mainly assesses the mucosal surface (11,230, 231,317,330). CT may change management in almost 15% of laryngeal cancer patients sent for preoperative imaging, usually by depicting cartilage involvement, extralaryngeal spread, and nodal disease (6). Compared with endoscopy, CT alone understages approximately 10% of superficial lesions; however, CT may upstage more than 20% of deep lesions (15,61). Additional pitfalls of CT include the inability to reliably differentiate inflammation and edema from tumor, to accurately identify subsite involvement in the presence of anatomic distortion from large tumors, and to clearly define tumor margins in the absence of welldeveloped fat planes (307,346). MRI has an added advantage of multiplanar display that can be helpful in establishing the extent of disease (56,125,132,222,447). The coronal display is especially useful in determining subglottic extension, and the sagittal display is equally helpful in identifying pre-epiglottic spread (125,212). Tumors are generally intermediate signal intensity on T1-weighted and proton density— weighted images and hyperintense on T2-weighted images. Lesions tend to moderately enhance with gadolinium (319). T1-weighted images readily assess tumor invasion in fat, whereas T2-weighted images are useful for detecting tumor invasion of muscle and soft tissues. As with CT, MRI tends to underestimate superficial disease, but is more accurate with higher staged lesions (55). Differentiating edema from tumor is difficult with MRI, but MRI is generally superior to CT in detection of cartilage invasion (55,56,125,412). Together, radiologic and clinical findings provide essential information on the staging of tumors, which in turn directs appropriate therapy (224,276–278).

Supraglottic Carcinoma The supraglottis is the superiormost division of the larynx and extends from the tip of the epiglottis to the level of an imaginary transverse plane that bisects the lateral apex of the laryngeal ventricle (69,148). The supraglottic region contains the suprahyoid and infrahyoid portions of the epiglottis, the false vocal cords, the aryepiglottic folds, the preepiglottic space, and the lateral paralaryngeal spaces (224). Almost 30% of laryngeal cancers occur in the supraglottis (148,423). Lesions produce few symptoms until the tumor is quite advanced, and more than half of these patients present with metastatic lymphadenopathy (Figs. 4-46 and 4-47) (148). Supraglottic cancers are classified as anterior supraglottic, arising in the epiglottis, and posterolateral supraglottic, which are further subdivided into marginal and false vocal cord/ laryngeal ventricle lesions (345). Anterior supraglottic lesions often grow circumferentially. Anterior inferior epiglottic tumors may invade the preepiglottic space and eventually the base of the tongue (Figs. 4-46 and 4-48) (50,224,291).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 174

174

Chapter 4

B

A Figure 4-46 Supraglottic squamous cell carcinoma. Enhanced computed tomography (A) demonstrates a large necrotic epiglottic mass (arrows) completely filling the preepiglottic space and compromising the airway. Image (B), more caudal to the previous slice, reveals a markedly thickened epiglottis and aryepiglottic folds (arrows) infiltrated with tumor. A metastatic lymph node is also identified (asterisk).

Marginal lesions arise on the medial surface of the aryepiglottic folds and may grow exophytically (Fig. 4-49). Lesions may invade the infrahyoid epiglottis, the false vocal cords, or cricoarytenoid region, or even the paralaryngeal space and pyriform sinuses (Fig. 4-50). False vocal cord/laryngeal ventricle lesions grow submucosally to invade the paralaryngeal space (Fig. 4-51) (26,50,148,224).

The thyroid cartilage is usually uninvolved until the tumor has grown into neighboring anatomic regions (50,291). Imaging is extremely useful in staging supraglottic carcinoma and may discover more advanced disease than clinically suspected in up to 20% of patients (89). Additionally, quantitative methods may prove valuable. For example, in patients with supraglottic carcinoma, studies suggest pretreatment CT-derived tumor volumes may be predictive of local control following radiation or surgery (164,235,263). Both CT and MRI define the primary tumor well because of the surrounding air and fat. Imaging assists in establishing that a 3-mm to 5-mm tumor-free margin is present above the anterior commissure, necessary for conservation

Figure 4-47 Supraglottic carcinoma. Enhanced computed tomography demonstrates a soft tissue attenuation mass (m) arising from the epiglottis and encroaching on the preepiglottic (black arrow) and paralaryngeal (double black arrows) spaces and right aryepiglottic fold (white arrow). Metastatic lymph nodes (asterisks) are also evident.

Figure 4-48 Supraglottic carcinoma. Large, predominantly hypodense, anterior epiglottic mass (arrows) fills the pre-epiglottic space at the level of the hyoid bone (H). The airway (white asterisk) is compressed by the lesion.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 175

Neck

175

Figure 4-49 Supraglottic carcinoma. Tumor thickens the left side of the epiglottis and extends into the paralaryngeal space and left aryepiglottic fold (arrows); normal right aryepiglottic fold (arrowhead ).

Figure 4-51 Localized false vocal cord tumor. Carcinoma involving the left false vocal cord (white arrows) obliterates the normal low-attenuation paralaryngeal space and bulges into the airway; normal right paralaryngeal space (arrowheads).

surgery (50,224). Moreover, imaging is essential in evaluating areas concealed from the view of the endoscopist. Pre-epiglottic fat invasion is demonstrated on both CT and MR images; however, axial and sagittal T1-weighted MR images show pre-epiglottic infiltration to the best advantage (57,217,291). Care must be taken not to misdiagnose the normal hyoepiglottic ligament as cancer (see Fig. 4-41) (50). Imaging also evaluates the paralaryngeal space (Fig. 4-50). Soft tissue infiltration of the normal fat-containing paralaryngeal space implies tumor invasion (see Figs. 4-47, 4-49, and 4-50) (224). In large tumors, MRI is superior in demonstrating cartilage involvement.

(50,180,345,423). Almost 75% involve the anterior half of the true cords (50,140). Patients usually present early in the disease process because of hoarseness. Anterior spread jeopardizes the anterior commissure. Soft tissue thickening of the anterior commissure greater than 1 mm in close apposition to a tumor on imaging implies tumor invasion, although the anterior commissure may be greater than 1 mm in thickness in more than 40% of normal patients (50,148,191,224). Air adjacent to the inner surface of the thyroid prominence effectively excludes cancer involvement. Once the anterior commissure is affected, further growth threatens the supraglottic region, the contralateral cord, and the thyroid cartilage (Fig. 4-52) (50,52). Posterior spread threatens the posterior commissure,

Glottic Carcinoma The glottis is subjacent to the supraglottis and extends 1 cm below the apex of the laryngeal ventricle. The true vocal cords dominate this anatomic region (69,71). Approximately 50% to 70% of laryngeal cancers involve the glottis

Figure 4-50 Supraglottic carcinoma. Computed tomography scan demonstrates a soft tissue lesion filling the normally fatcontaining left paralaryngeal space (arrowheads) and abutting the left thyroid lamina (arrows).

Figure 4-52 Carcinoma of the left true vocal cord. Enhanced computed tomography in a patient with a clinically fixed left true vocal cord. An enhancing mass (arrowheads) of the anterior left true cord extends laterally into the paraglottic space to the thyroid cartilage and anteriorly to involve the anterior commissure (arrow). The anterior right true cord is also involved. The thyroid cartilage is intact and not invaded by tumor. Laryngeal prominence (P).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 176

176

Chapter 4

Figure 4-53 Carcinoma of the true vocal cord with involvement of anterior and posterior commissures. Left true vocal cord tumor extends posteromedially over the arytenoid cartilage toward the posterior commissure (arrows) and anteriorly across the anterior commissure (arrowhead ).

True cord lesions superficial to the conus elasticus usually do not interfere with normal cord mobility. Limited cord motion may result from the bulk of larger tumors or invasion of the thyroarytenoid muscle. Complete fixation of a cord is most commonly the result of replacement of the thyroarytenoid muscle by invasive tumor (197,207,291). Only rarely does invasion of the cricoarytenoid joint as a mechanical cause for cord fixation occur without concomitant thyroarytenoid muscle involvement (197). Cord fixation also may be caused by subglottic extension or adherence of tumor to the thyroid cartilage (50). Because small glottic tumors present early and are well evaluated with endoscopy, imaging is rarely necessary for such T1 lesions (203). When performed, imaging may demonstrate no abnormality or may show subtle asymmetry of the cords (Fig. 4-55) (89). For larger lesions, coronal MRI is useful for demonstrating the relation of the tumor to the ventricle and subglottic space (222). Causes for a false–positive diagnosis of carcinoma include benign polyps and granulomatous disease (307).

Subglottic Carcinoma the arytenoids, and the cricoarytenoid joint (Fig. 4-53). Involvement of the arytenoids may rotate or displace these cartilages (442). Invasion of the posterior commissure further endangers the contralateral cord and eventually the soft tissues of the neck (417). Glottic tumors may extend superiorly to involve the paralaryngeal space and supraglottis or inferiorly to involve the subglottis (26,224). Inferior extension greater than 8 to 9 mm anteriorly and 5 to 6 mm posteriorly usually requires total laryngectomy (Fig. 4-54) (197,277). Assessment of cord mobility is essential for cancer staging and is best accomplished by endoscopy (224). Further evaluation of the impaired cord requires imaging (237).

The subglottis (infraglottis) is immediately subjacent to the glottis and extends to the inferior margin of the cricoid cartilage (322). Primary subglottic cancers are rare and represent only 2% to 6% of laryngeal carcinomas (180). More commonly, subglottic tumors represent extensions from the glottic region or pyriform sinus (Figs. 4-54 and 4-56) (50). Subglottic extension from glottic primary cancer is usually associated with cord fixation (197). Patients often present with dyspnea, stridor, and pain; over 40% have cervical lymphadenopathy at initial examination (194,322). Subglottic carcinoma tends to spread circumferentially and is initially limited by the conus elasticus (322). The relation of the tumor to the cricoid cartilage is critical since

B

A Figure 4-54 Carcinoma of the left true vocal cord with subglottic extension. Enhanced computed tomography (CT) (A) demonstrates a smooth mass (arrow) of the left true vocal cord projecting into the airway. CT scan at the level of the cricoid (B) reveals inferior extension of the tumor (arrowhead ) , causing thickening of the mucosa.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 177

Neck

177

eventually involve the true cords. The thyroid gland and hypopharynx also are at risk (180). Subglottic cancer is not easily evaluated by endoscopy, and imaging is therefore necessary to accurately stage these tumors. On CT and MR images, no measurable soft tissue thickness should be demonstrated between the cricoid cartilage and the airway. Any soft tissue thickening or exophytic tissue should be considered carcinoma (180). MRI has the additional advantage of scanning in the coronal plane, which allows tracing of the tumor below the conus elasticus (69).

Transglottic Carcinoma

Figure 4-55 Glottic carcinoma. Enhanced computed tomography at the level of the vocal processes of the arytenoids (arrowheads) reveals nodular irregularity of the true vocal cords (white arrows). The lesion is slightly hyperdense. Diagnosis of small lesions relies heavily on the abnormal contour of the airway caused by the mass. Interpretation also depends on detailed clinical information. Benign polyps may have an identical appearance.

the cricoid cartilage must be intact for successful conservative surgery. Almost 50% of subglottic carcinomas have cricoid or thyroid cartilage invasion at clinical presentation (197). Inferior spread may involve the trachea (26). Postsurgical tracheotomy stomas are especially at risk for subglottic cancer recurrence because of involvement of the paratracheal lymph nodes (249). Superior spread may

Transglottic cancers are defined as lesions that involve the true vocal cords, laryngeal ventricle, and at least the supraglottis and/or the subglottis (50). These tumors are often associated with vocal cord fixation, deep invasion, and metastatic lymphadenopathy; transglottic tumors exceeding 2 cm in diameter frequently involve the cartilage (197). The extent of these lesions may not be fully appreciated clinically due to submucosal spread (Fig. 4-57) (408).

Hypopharyngeal Carcinoma Hypopharyngeal carcinomas are about a third as common as tumors in the larynx (2,443). These aggressive cancers arise from the pyriform sinuses (60%), the postcricoid region (25%), and the posterior pharyngeal wall (15%) (2,46,148). Early symptoms are minimal, and the vast majority of patients present with advanced disease. Factors contributing to the poor prognosis are submucosal invasion of the esophagus and extensive lymphadenopathy resulting from the adjacent abundant lymphatic network. Almost 80% of patients have palpable lymph nodes at

A

B Figure 4-56 Carcinoma of the true vocal cords with subglottic extension. Enhanced computed tomography (A) demonstrates thickening of the anterior commissure (arrow) by a mass involving both true vocal cords. B: The mass descends into the subglottic region and perforates the cricothyroid membrane and inferior thyroid cartilage to invade the soft tissues of the neck (black arrowheads). Portions of the mass are hypodense as a result of necrosis. The airway (asterisk) is markedly narrowed. Prior radiation therapy has caused skin thickening (white arrowheads).

5063_Lee_Ch04pp0145-0224 10/13/05 4:50 PM Page 178

178

Chapter 4

B

A Figure 4-57 Transglottic carcinoma. Enhanced transaxial T1-weighted magnetic resonance image (A) reveals a large supraglottic soft tissue mass (arrowheads) that invades the left paralaryngeal space, the left thyroid cartilage lamina, and paralaryngeal muscles (arrow). Sagittal T1-weighted image (B) demonstrates the full extent of tumor (arrowheads) as it infiltrates to the level of the true cords.

presentation (2). Moreover, patients have a 20% chance of developing a second primary malignancy within 5 years of diagnosis (2). As in primary laryngeal carcinoma, tobacco and alcohol are important risk factors. Additionally, PlummerVinson syndrome also confers higher risk. This syndrome, affecting middle-aged women, consists of dysphagia, hypopharyngeal and esophageal webs, and iron deficiency anemia. Cancers usually develop in the postcricoid region proximal to the webs and are thought to be the result of stasis and irritation of the mucosa (2). Pyriform sinus lesions usually spread by direct invasion of adjacent structures and by lymphatic infiltration to regional nodes (Fig. 4-58) (197,215,443). Cancers arising in the lateral wall usually spread posterolaterally and may eventually involve the carotid artery, thyrohyoid membrane, and thyroid cartilage (Fig. 4-59) (50,81,324). Tumors originating in the medial wall may invade the aryepiglottic folds, cricoid cartilage, cricoarytenoid joint, arytenoid cartilage, and paralaryngeal space, mimicking a primary laryngeal process (50,197,215). Large tumors may widen the cricothyroid space and spill into the postcricoid region or into the soft tissues of the neck (Fig. 4-60) (317,324,423). Masses that have approached to within 5 mm of the esophageal verge on imaging usually are found to involve the verge at surgery (324). Superior and anterior spread may involve the preepiglottic space (443). Postcricoid region tumors are difficult to diagnose because the postcricoid hypopharynx is usually collapsed. Tumors creep behind the cricoid cartilage and may then bulge posterolaterally into the soft tissues of the neck. Posterior hypopharyngeal masses thicken the retropharyngeal space and may invade the retropharyngeal lymph nodes (Fig. 4-61) (73,74). Inferiorly, tumors may extend to the cricopharyngeus muscle.

Special Issues in Cancer of the Larynx and Hypopharynx The laryngeal cartilage integrity is an important feature of cancer staging. The presence of cartilage invasion implies a T4 lesion, which precludes conservative surgery (26,268,399). Moreover, irradiation of invaded cartilage may predispose to perichondritis and cartilage necrosis (406). The thyroid cartilage is most commonly affected. Sites of invasion within the thyroid cartilage include the anterior lamina, posterior border, attachments of the cricothyroid membrane, anterior commissure ligament, and points of vessel

Figure 4-58 Pyriform sinus carcinoma. Dense soft tissue mass (arrowheads) partially effacing right pyriform sinus and infiltrating paralaryngeal space on enhanced computed tomography image. The mass approaches the right carotid artery (C ) , but does not involve the vessel. Normal left pyriform sinus (asterisk), normal left paralaryngeal space (p).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 179

179

Neck

A

B

Figure 4-59 Pyriform sinus carcinoma, with thyroid cartilage destruction and extralaryngeal extension. A: On T1-weighted magnetic resonance image, a large hypointense tumor (t) is seen in the region of the right pyriform sinus. Fat within the medullary cavity in the posterior portion of the left thyroid lamina (small arrows) is high in signal intensity, whereas the right thyroid lamina is destroyed and its medullary cavity is replaced by low-intensity tumor (arrowheads). Distinction between the tumor and strap muscles is poor. The tumor abuts the carotid artery (C) but does not involve its wall; internal jugular vein (J ) . B: On T2-weighted image, the extralaryngeal extension of hyperintense tumor (large arrow) and the strap muscles (S ) are better delineated. Contrast between the tumor and fat within the paralaryngeal space is decreased; carotid artery (C); jugular vein (J ) ; sternocleidomastoid muscle (SCM).

penetration (Fig. 4-62) (249,268). The cricoid cartilage may be involved at its base, at attachments of the cricotracheal ligaments, and at the cricoarytenoid joint (268,346). Generally, cartilage invasion is related to tumor size and location. Tumors originating below the apex of the arytenoid cartilage and exceeding 16 mm in size have a high likelihood of cartilage invasion (13,399). Cancers in close proximity to cartilage, such as subglottic, transglottic, and pyriform sinus tumors, have a greater than 50% chance of associated cricoid and lower thyroid cartilage involvement (50,208,399). In contradistinction, suprahyoid supraglottic lesions involve cartilage much less frequently. No relationship between histologic cell type or grade of tumor

and probability of cartilage invasion has been described (268). Ossified portions of cartilage are at higher risk for invasion possibly because of vascular channel penetration (268). Intact perichondrium surrounding avascular unossified cartilage generally resists tumor encroachment, although newer studies suggest that invasion of nonossified cartilage may occur more frequently than previously suspected (26,268,291,346,432). Sclerosis of cartilage is one of the most sensitive indicators of cartilage involvement with tumor (28). Markedly sclerotic cartilages associated with an adjacent mass demonstrate actual invasion in almost 50% of cases at surgery (Fig. 4-63) (268,396). The process of cartilage involvement

A

B Figure 4-60 Pyriform sinus hypopharyngeal squamous cell carcinoma. Lateral view from barium esophagram (A) demonstrates a large mucosal based hypopharyngeal mass (arrows). Enhanced computed tomography (B) reveals a large hypopharyngeal mass (small arrowheads) extending posterolaterally to abut the left carotid artery (c) and displacing the barium-filled hypopharynx to the right (arrow). A small left jugular lymph node is noted (large arrowhead ).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 180

180

Chapter 4

Figure 4-61 Hypopharyngeal carcinoma. Computed tomography scan demonstrates a large hypopharyngeal carcinoma (arrowheads) that has grown posteriorly to involve the retropharyngeal space.

appears to progress in several stages. Initially, cartilage adjacent to tumor develops inflammatory changes, probably in response to humoral products released by the tumor. These inflammatory changes incite new bone formation within the cartilage prior to frank tumor invasion. Prostaglandins and interleukin-1 released by the tumor cells promote osteoclast activity and osteolysis. Finally, tumor penetrates the dissolved bony matrix (26). Nonmalignant causes of thyroid cartilage sclerosis include rheumatoid arthritis and polychondritis; these conditions usually can be differentiated from cancer by the presence

Figure 4-62 Carcinoma of the right true vocal cord with thyroid cartilage destruction. Mass involving the right true vocal cord invades the anterior commissure and destroys the right thyroid lamina (black arrowheads), and extends into the adjacent soft tissues (white arrowheads). Cortical thinning of the posterior aspect of the left thyroid lamina (arrows) is a normal variation and should not be interpreted as cartilage destruction.

Figure 4-63 Transglottic carcinoma, arytenoid cartilage involvement. Computed tomography scan shows a mass in the left vocal cord that extends to the thyroid cartilage (asterisk) and the left arytenoid cartilage (arrowhead ) . The mass escapes into the neck through the thyroarytenoid space (arrows). The left arytenoid cartilage is sclerotic compared with its companion on the right. This extensive mass also involved the supraglottic region.

of sore throat, difficult inspiration, subluxation of the arytenoid cartilages, or the absence of a mass (268). Frank destruction of cartilage by a mass and demonstration of intramedullary cancer are the most reliable imaging signs of cartilage invasion (29) (see Fig. 4-59). Because hyaline cartilage calcifies and ossifies in an irregular pattern, recognition of pathologic sclerosis or subtle invasion in the thyroid cartilage is difficult on CT images (see Fig. 4-39) (12,168,225,329,346,448). Because of its usual symmetry, sclerosis or destruction of the cricoid is easier to appreciate on CT images (Fig. 4-64) (50). MRI detects cartilage invasion better than CT does (30,52,166,412). Replacement of the normal hyperintense fatty marrow signal with isointense soft tissue signal on T1-weighted images indicates invasion (180). On proton density–weighted and T2-weighted images, tumor produces hyperintense signal (see Fig. 4-59) (291,412). Unfortunately, MR signal changes are nonspecific and cannot differentiate neoplasm from nonneoplastic inflammation (30). A mass abutting or surrounding cartilage must also prompt consideration of invasion, especially if the normally calcified outer rim of the cartilage is interrupted (180,224). Almost 10% of normal thyroid cartilages may have focal areas of altered signal intensity; however, an associated mass increases confidence in the diagnosis, especially if the mass exceeds 5 mL in volume (58,180). Large tumors expand and fragment the cartilage. Cartilaginous calcifications may lie outside the contour of the normal cartilage. Differentiating cancer involvement from primary laryngeal chondroma or chondrosarcoma is the predominantly noncalcified soft tissue mass of carcinoma compared with the often calcified matrix of cartilage-based lesions (307).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 181

Neck

181

Figure 4-64 Transglottic carcinoma with cartilage invasion. Computed tomography scan reveals a soft tissue mass (white arrows) thickening the mucosa and projecting into the airway at the level of the cricoid cartilage. This mass had spread from a glottic site and descended to the inferior margin of the cricoid cartilage. Note the cricoid sclerosis (arrowheads) and destruction, with marrow replacement by tumor (black arrow), that marked the cartilage invasion.

Carotid artery invasion presents an additional issue of clinical concern. Although relatively uncommon, carotid artery involvement is identified in 5% to 20% of patients undergoing radical neck surgery. Preoperative clinical diagnosis is difficult. More than 60% of patients with presumed invasion may not demonstrate invasion on pathologic examination. Carotid resection is dangerous. Carotid ligation is associated with a perioperative morbidity of cerebral infarction (13% to 80%), rupture, abscess, or fistula. The 1-year survival rate after carotid resection is 0% to 44% (454). Ultrasound may be useful in demonstrating hypoechoic changes in the invaded wall, but has the disadvantage of being operator dependent and limited by calcium within the vessel (141). Sonographic demonstration of malignant lymphadenopathy adherent to the carotid for 4 cm or greater in length is associated with an 80% chance of invasion (140). MRI is superior to CT in assessing carotid invasion. Circumferential attachment of tumor to 270 degrees or greater of the carotid diameter implies carotid involvement with a sensitivity approaching 100% (Fig. 4-65). Less than 180 degrees of circumferential contact with tumor usually implies a free carotid (454). Pitfalls in the MR interpretation of carotid invasion include image degradation resulting from patient motion or vascular motion, exaggeration of adventitial signal on enhanced T1-weighted images, previous radiation therapy, asymmetric wall thickening resulting from atherosclerosis, and laminar flow along vessel walls (454). The suggestion of vascular invasion on imaging in a patient requiring extensive neck surgery should prompt a preoperative trial balloon occlusion with catheter angiography (176). For cancers that have eroded

Figure 4-65 Cervical recurrence, floor of the mouth carcinoma. A contrast-enhancing soft tissue mass (white arrowheads) is identified in the left neck at the level of the thyroid cartilage (arrows). The carotid artery (black arrowhead ) is encased by the lesion. The subcutaneous fat and endolarynx are edematous because of previous radiation therapy. Note that the hyoid bone (H) and thyroid cartilage appear on the same imaging slice due to contraction of the thyrohyoid membrane.

branches of the external carotid artery and have produced catastrophic hemorrhage, angiographic embolization is effective (449). Differentiating pyriform sinus cancer from supraglottic cancer can be difficult, especially in bulky lesions. Prognostic and therapeutic implications differ between the two sites of origin. Pyriform sinus lesions tend to be unilateral and are more likely to invade lymph nodes. Supraglottic lesions tend to spread circumferentially and then superiorly or inferiorly. Multiple primary cancers develop in 15% to 30% of patients who have been otherwise successfully treated for their original primary head and neck tumors (40,50,51). More than 5% of patients develop a lung cancer within 12 years (51). These tumors are usually metachronous rather than synchronous and most often involve the upper aerodigestive tract. The risk factors of tobacco and alcohol abuse undoubtedly contribute to this association. Periodic chest radiography and barium esophagography remain important and cost-effective tools for screening second primary lesions (51); bronchoscopy with cytological washings may be more valuable in diagnosing radiographically occult lesions (51). Distant metastases are usually manifested late in the course of the disease. With more effective local and regional treatments, metastases may be seen in lungs, bone, and liver (52). Radionuclide bone scintigraphy is an excellent screening tool.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 182

182

Chapter 4

Benign Mucosal Disease of the Larynx and Hypopharynx

Submucosal Disease of the Larynx and Hypopharynx

Benign mucosal disease of the larynx and hypopharynx is commonly encountered in the typical otolaryngologic office practice. Easily recognizable lesions, such as acute and chronic laryngitis, are treated in an outpatient setting, and suspicious masses are readily biopsied without the need for imaging. Croup is an inflammation of the subglottic larynx caused by parainfluenza virus type 1. It typically occurs in young children. Radiographic studies are usually not indicated, but are useful to exclude other causes of stridor, such as a foreign body. Symmetric subglottic airway narrowing or “penciling” of the airway on frontal plain radiographs of the neck is caused by inflammatory edema (47,65,71,90,91,208,442). Membranous or bacterial croup is characterized by diffuse inflammation of the larynx, trachea, and bronchi with adherent exudate and mucus on the surface of the upper tracheal mucosa (442). Epiglottitis is an inflammation caused by Hemophilus influenzae and involves the epiglottis and aryepiglottic folds. These normally well-defined structures become edematous and enlarged, resulting in a rounded, thumblike appearing epiglottis on lateral plain radiographs. The hypopharynx and pyriform sinuses usually are mildly to moderately overdistended. Foreign bodies, burns, epiglottic cysts, neoplasms, granulomatous disease, and angioneurotic edema may also enlarge the epiglottis (242,351,442). Cross-sectional imaging is not necessary in croup and epiglottitis. Laryngeal tuberculosis produces nonspecific bilateral diffuse vocal cord thickening and may be accompanied by thickening of the epiglottis (60%), edema of the paraglottic spaces (67%), increased density of the pre-epiglottic space (50%), thickening of the glossoepiglottic folds (25%), and cervical lymphadenopathy (40%). The laryngeal cartilages usually are uninvolved. The diagnosis should be suspected in patients with laryngeal disease and pulmonary tuberculosis (253). Benign neoplasms are additional causes for aerodigestive tract lesions. An obstructed minor salivary gland may produce a small cystic lesion in the vallecula or epiglottic surface that is incidentally discovered on imaging (67). Vocal cord polyps may occur after abuse of the voice. Imaging is rarely requested but may show a nodular contour deformity indistinguishable from a small cancer. Over 80% of benign mucosal tumors are papillomas (432). Papillomas are the most common pediatric laryngeal tumor (406). These lesions are usually multiple and may appear as nodular excrescences on the false and true cords. Imaging is nonspecific and may demonstrate nodular contour abnormalities on the mucosal surfaces projecting into the airway.

Submucosal masses of the larynx and hypopharynx are decidedly rarer than mucosally derived carcinoma. To the endoscopist, submucosal lesions usually present as masses that are covered with intact mucosa and that displace normal structures. Although cancer is considered a mucosal lesion, over 50% of submucosal masses are eventually determined to be lateral laryngeal ventricle or saccule cancers that have invaded deeply into the paralaryngeal space (323). Laryngoceles represent over 20% of the true submucosal lesions of the larynx (323). These lesions consist of airfilled diverticula arising from the saccule of the laryngeal ventricle and may be mistaken clinically for a submucosal neoplasm (131,355,416). Laryngoceles within the paralaryngeal space and confined within the thyroid lamina are termed internal, whereas lesions that pierce the thyrohyoid membrane to present in the lateral neck are termed external. Most are a combination of external and internal and are referred to as mixed (Fig. 4-66). An external lesion does not occur without an internal component (355). Almost 25% are bilateral (423). Laryngoceles are usually asymptomatic, but have been associated with airway obstruction, pyoceles, and vocal cord paralysis (139). The lesions themselves are overwhelmingly benign, but a small ventricular cancer can obstruct the saccule, producing a distal, fluid-filled laryngocele, termed a laryngeal mucocele (138,355,406). Laryngoceles are often incidental findings on imaging and present as air-filled or occasionally fluid-filled cavities within the paralaryngeal space. Rarely, cystic lesions such as cystadenomas can mimic a laryngeal mucocele (108). Hemangiomas may be subdivided into infantile type and adult type (27). Subglottic infantile hemangioma is the most common laryngeal and upper tracheal neoplasm

Figure 4-66 Mixed laryngocele. Enhanced computed tomography reveals an air-filled laryngocele straddling the thyrohyoid membrane. The internal component (arrowhead ) is medial to the hyoid bone (asterisk), and the external component (arrow) is lateral to the hyoid.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 183

Neck

in the newborn and young infant. The lesion typically appears as a well-defined mass in the posterior or lateral portion of the subglottic airway. Although the subglottic narrowing is usually eccentric, circumferential narrowing suggestive of croup may be seen (351,442). Infantile hemangiomas may regress spontaneously. The adult cavernous form may contain phleboliths. On imaging, all types enhance vigorously (27). Differential diagnostic considerations of enhancing, highly vascularized, submucosal laryngeal masses include laryngeal paragangliomas and metastases (27). Cartilage-derived lesions are rare and account for less than 10% of laryngeal submucosal lesions (323). Chondrosarcomas of the larynx are slowly growing malignant neoplasms that comprise approximately 70% of cartilage tissue lesions. Most patients present in the sixth and seventh decades of life. Symptoms include hoarseness, dyspnea, and dysphagia. Approximately 80% of these lesions arise in the cricoid cartilage, followed in frequency by the thyroid cartilage. Lesions in virtually all patients demonstrate coarse or stippled calcifications (Fig. 4-67) (27, 444). Features differentiating chondrosarcoma from cancer include the generally older age at diagnosis, absence of smoking history, and predominately calcified tumor matrix. Chondromas are benign lesions that arise from either hyaline or elastic cartilage or other soft tissues of the neck. Presentation is usually decades earlier and tumor volume is generally smaller than with its malignant counterpart, chondrosarcoma (78). Unfortunately, radiologic distinction is extremely difficult (444). Chondrometaplasia is a benign condition in which nodules of cartilage arise in the laryngeal soft tissues. This condition is usually asymptomatic if the spicules of cartilage remain 2 mm or less in

183

size (23). Lesions arising in the vicinity of the cricoid or thyroid cartilages may be difficult to differentiate from chondromas or chondrosarcomas on imaging. Schwannomas of the larynx are nerve sheath tumors that typically involve the sensory nerves, such as the internal branch of the superior laryngeal nerve (423). This lesion usually presents in the fourth to sixth decades of life as a palpable mass (292). Almost 15% of solitary nerve sheath tumors are associated with malignant lesions at other sites (72). On imaging, schwannomas tend to be well delineated and hyperintense on T2-weighted images and isointense to hypointense on T1-weighted images. Enhancement is homogeneous, although cystic degeneration may be demonstrated (292). Benign and malignant minor salivary gland tumors are rare causes of submucosal masses of the larynx and hypopharynx. Pleomorphic adenomas, mucoepidermoid carcinomas, adenocarcinomas, and adenoid cystic carcinomas have been reported (27). The latter lesion may have a hypodense center on a CT image, but is otherwise indistinguishable from squamous cell carcinoma and should be considered in a patient with a laryngeal mass and no history of smoking or drinking (75). Other rare primary submucosal masses and conditions include paragangliomas, atypical carcinoid tumors, fibrosarcomas, liposarcomas, lymphomas, and amyloidosis (Fig. 4-68) (27,100,101,177,323). Metastases to the larynx usually occur in the terminal stages of widely disseminated malignancy. Primary tumors include melanoma (30%), renal carcinoma (15%), lung carcinoma (10%), breast carcinoma (10%), and prostate carcinoma (5%) (99). Sites of tumor deposit include the supraglottis (40%), subglottis (20%), glottis (5%), and multiple regions (35%) (99). Metastases usually begin as submucosal masses, but may secondarily ulcerate the mucosa. Patients may complain of hoarseness, dysphagia, or hemoptysis.

Figure 4-67 Chondrosarcoma of the cricoid cartilage. Computed tomography demonstrates stippled calcification (arrowheads) within a large mass (small arrows) arising from the cricoid cartilage and extending into the extralaryngeal tissues. The thyroid cartilage (large arrows) is displaced anteriorly and to the left. (From Wippold FJ 2nd, Smirniotopoulos JG, Moran CJ, et al. Chondrosarcoma of the larynx: CT features. AJNR Am J Neuroradiol 1993;14:453–459, with permission.)

Figure 4-68 Lymphoma of the hypopharynx. A predominately submucosal right pyriform sinus mass (black asterisk) invades the preepiglottic and right paralaryngeal spaces (arrowheads), with effacement of the right pyriform sinus. Biopsy confirmed a diffuse, large, B-cell lymphoma. Left pyriform sinus (white asterisk).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 184

184

Chapter 4

Imaging usually is nonspecific, and diagnosis is established with biopsy.

Other Conditions of the Larynx and Hypopharynx Paralysis of the vocal cord, in the absence of an intrinsic laryngeal lesion, may be the result of any process involving the vagus nerve or its recurrent laryngeal branch between the jugular foramen and its entrance into the larynx. Almost 75% of such patients have unilateral paralysis (403). Almost 90% of these paralyses are the result of peripheral nerve lesions, and only 10% are central in origin; the latter typically are accompanied by other cranial neuropathies (3). The peripheral location of the vast majority of lesions reflects the long and redundant routes of the recurrent laryngeal nerves. Specific etiologies include neoplasm (36%), postoperative complication such as from parathyroid and thyroid surgery (25%), inflammation (13%), and idiopathic (403). Congenital central nervous system anomalies are associated with childhood vocal cord paralysis (39). Although laryngeal cancers can paralyze the cord, direct infiltration of the recurrent or superior laryngeal nerves by local tumor is very rare (249). Radiologic evaluation of the isolated paralyzed cord should entail imaging of the entire vagus nerve course from the skull base to the pulmonary hila. CT is excellent for evaluating the neck and chest; MRI is superior for the skull base (129,182,403). CT signs of cord paralysis include paramedian position of the cords, displaced arytenoid cartilage, ipsilateral dilatation of the pyriform sinus, tilting of the thyroid cartilage, and prominent laryngeal ventricle (3). Laryngomalacia or congenital flaccid larynx causes inspiratory stridor in the neonate and young infant (91). This condition usually resolves by 1 year of age. Laryngoscopy demonstrates flaccidity of the epiglottis, aryepiglottic folds, or the entire larynx, which collapses during inspiration (442). Laryngotracheal stenosis is a nonneoplastic condition characterized by focal or diffuse airway narrowing. Diffuse narrowing may be caused by tuberculosis, sarcoidosis, histoplasmosis, relapsing polychondritis, amyloidosis, Wegener granulomatosis, or ulcerative colitis with tracheobronchitis (Fig. 4-69). Focal stenosis may be seen following endotracheal or tracheostomy tube placement or neck trauma (35). Imaging reveals a narrowed laryngeal, subglottic, or tracheal airway; the contour may be smooth, lobulated, tapered, or irregular (35). Laryngeal trauma usually results from direct blunt injury to the anterior neck. The usual mechanism for the observed damage is the forceful impact of the larynx against the cervical spine. CT is the preferred imaging modality in the acutely injured patient because of its speed and ability to define calcified cartilages and soft tissue damage (Fig. 4-70 to 4-72) (121,232). Spiral techniques are especially well suited for this situation. The thyroid cartilage may be fractured in almost 50% of cases (see Figs. 4-70 and 4-71). The cricoarytenoid joint also may be dislocated. Traumatic intubation also may cause severe soft tissue damage, soft tissue emphy-

Figure 4-69 Diphtheria. Enhanced computed tomography scan in a patient with a childhood history of severe diphtheria and a chronic tracheotomy. The laryngeal cartilages are collapsed and the laryngeal airway occluded. The thyroid (arrowheads) and cricoid (arrow) cartilages are misshapen. The distance between the thyroid and cricoid cartilage is diminished.

sema, and arytenoid cartilage dislocation. The cricoid cartilage also may fracture (see Fig. 4-72). Because of its ring configuration, usually two or more sites are involved. Rarely, the epiglottis may be disrupted (87). Hemorrhage and edema may spread into the paraglottic and pre-epiglottic spaces (428). Nondisplaced fractures may be treated conservatively; however, displaced and comminuted fractures usually require surgical intervention (32,328).

Thyroid The thyroid gland occupies the visceral space and consists of bilateral lobes that lie adjacent to the airway. The upper poles ascend to the level of the thyroid cartilage and the body of the gland lies at the level of the cricoid cartilage and trachea. A midline isthmus of tissue joins the lobes and straddles the anterior trachea (295). On CT images,

Figure 4-70 Fracture of thyroid cartilage. A minimally depressed fracture of the left thyroid lamina (arrowhead) is seen. The adjacent strap muscles are slightly thickened secondary to hemorrhage or edema. The true vocal cords and cricoid cartilage are normal.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 185

Neck

Figure 4-71 Comminuted fracture of thyroid cartilage. Multiple fractures of the thyroid cartilage are seen, with extensive hemorrhage and edema severely compressing the airway (arrow).

the gland generally is hyperdense compared with muscle because of its iodine content on noncontrast images; it dramatically enhances because of its vascularity post-contrast administration (121,178) (Fig. 4-73). CT attenuation may be diminished in patients on thyroid replacement therapy (121). The thyroid gland originates in the epithelium of the foramen cecum, a small pit located at the junction of the anterior two thirds and posterior one third of the tongue (295,316). This thyroid tissue descends ventral to the foregut during embryologic development and eventually comes to rest anterior to the trachea. Persistent tissue can be found anywhere along the path of migration. Lingual thyroid tissue, which makes up 90% of ectopic thyroid tissue (439), represents residual thyroid tissue located in the base of the tongue and presents as a well-defined soft tissue mass (143,192). Patients are generally asymptomatic, but may present with dysphagia, dysphonia, or dyspnea (439). On MR images, the tissue is hyperintense on both T1-weighted and T2weighted images compared with muscle, therefore assisting in the differentiation from invasive squamous cell cancer of the tongue. Lingual thyroid tissue is subject to all of the pathologic conditions associated with the normally located thyroid gland. Complete surgical removal must be tempered with the knowledge that more than 70% of patients

185

have no other functioning thyroid tissue (245). Radionuclide scintigraphy can determine the presence of any functioning gland in the neck prior to surgery (143). The differential considerations of ectopic thyroid tissue on imaging include epidermoid inclusion cysts, angiomas, fibromas, adenomas, hemangiomas, minor salivary gland tumors, and carcinoma (143,192). Thyroglossal duct cysts are cystic derivatives of residual thyroid tissue located along the migration pathway. The clinician is usually alerted to thyroid pathology by signs and symptoms of a hypofunctioning or hyperfunctioning gland or of a neck mass detected during physical examination. Issues of hormone production usually are investigated with a battery of sensitive serum assays. Imaging seldom is necessary. In certain circumstances, such as in the evaluation of neonatal hypothyroidism or in the differentiation of various causes of hyperthyroidism, imaging is usually accomplished with nuclear scintigraphy; CT and MRI have little role (105). Recent suggestions that MRI signal intensities may reflect thyroid function in certain conditions are intriguing, but not likely to alter management in the near future (60,389). Imaging is more useful in the evaluation of the enlarged gland. Goiters are benign enlargements of the thyroid gland that exceed 50 g in weight. Goiters may be seen in many conditions and are classified as simple or multinodular. Simple goiters arise in patients who are chronically deprived of dietary iodine and are not usually associated with hyperthyroidism. Multinodular goiters may become quite large and asymmetrically bulky and may cause hyperthyroidism. Such goiters have been associated with dietary factors, autoimmune disorders, and exposure to goitrogens (295). The CT appearance of goiters is variable, but the enlarged gland usually is inhomogeneous and often asymmetric (Fig. 4-74). Occasionally, some goiters have calcification, and virtually all enhance (304). Goiters may descend into the mediastinum, particularly the anterior compartment. CT signs suggesting a thyroid origin of a mediastinal mass include the intimate association of the superior pole of the mass with the thyroid gland, the hyperdensity of the lesion

A

B Figure 4-72 Fracture of cricoid cartilage. A: A fracture (arrow) through the right side of the cricoid cartilage is seen. B: Edema/hemorrhage thickens the right true vocal cord and narrows the airway (arrowheads). Extensive soft tissue air is also present.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 186

186

Chapter 4

A

B

Figure 4-73 Normal thyroid gland. Unenhanced computed to-

C

mography images through the upper portion (A), midportion (B), and lower portion (C ) of the thyroid gland demonstrate the two lobes of the gland (black arrowheads) as structures of relatively high attenuation value adjacent to the trachea (T ) . The thyroid isthmus (white arrowhead) connects the right and left lobes. Common carotid artery (C); internal jugular vein (J); esophagus (E); longus colli muscles (arrows).

compared with the surrounding tissue, the presence of calcifications, avid and persistent enhancement of the mass, and the close proximity of the lesion to the trachea (105,301). Generalized enlargement of the gland also may be caused by diffuse disease, such as Hashimoto thyroiditis, the

Figure 4-74 Goiter. Enhanced computed tomography reveals an enhancing heterogeneous soft tissue mass (arrows) that originated in the thyroid gland and descended through the thoracic inlet, deviating the trachea (asterisk) to the right.

most common chronic thyroid inflammation. It is mediated by an autoimmune mechanism and may be associated with non-Hodgkin lymphoma of the thyroid, collagen vascular disease, or Graves disease. Middle-aged women are more commonly affected (389,458), and many patients are hypothyroid. The gland is usually enlarged (387) and hyperintense on proton density–weighted MR images in untreated patients (389). Reidel thyroiditis is another chronic inflammatory process and is characterized by extensive fibrosis that may become bulky enough to compress the esophagus (451,458). Associations include mediastinal fibrosis, sclerosing cholangitis, and orbital pseudotumor. The gland is usually hypodense on CT images and hypointense on both T1-weighted and T2-weighted MR images (458). Graves disease, also known as diffuse toxic goiter, causes enlargement of the gland and is associated with characteristic orbital findings of proptosis, extraocular muscle enlargement that spares the tendinous insertions, and increased orbital fat. Small focal enlargements or solitary nodules of the thyroid are best evaluated with fine-needle aspiration. Scintigraphy may be useful in cases of nondiagnostic cytologies, family history of Hashimoto thyroiditis, or history of anticoagulant medication precluding needle aspiration. CT and MRI are usually not indicated (105). Thyroid cysts are

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 187

Neck

fairly common causes of palpable masses. Cysts may grow rapidly. Hyperintensity resulting from protein content or hemorrhage is frequently observed on T1-weighted images. Care must be taken when evaluating cervical carotid disease with time-of-flight MR angiography in patients with thyroid cysts in the vicinity of the artery. Hyperintensity of the cyst may be projected over the artery and mimic a pseudoaneurysm (7). Tumors of the thyroid gland consist of benign adenomas and malignant neoplasms. Risk factors for developing tumors include exposure to ionizing radiation during childhood and prolonged exposure to thyroid-stimulating hormone. Radiation may have been in the form of prior radiotherapy for neck or thymus lesions or result from environmental disasters, usually with a requisite latent period of decades. The relationship between the incidence of tumor occurrence and radiation dose is linear. Adenomas are benign masses that most frequently present clinically as slowly growing nodules, sometimes associated with hyperthyroidism. Usually discrete and solid, these lesions may undergo hemorrhage and cystic degeneration. Adenomas often are incidentally found on CT and MR scans performed for other indications, such as degenerative cervical spine disc disease. On CT images, the lesions are usually homogeneous and of a slightly different attenuation compared with the rest of the thyroid. Heterogeneous regions may result from hemorrhage (387). On MR images, adenomas may be of variable intensity on T1weighted and T2-weighted images, depending on protein content or hemorrhage. Cancer of the thyroid gland affects young and middleaged adults; women are almost twice as commonly affected. Most cancers are well differentiated. Papillary thyroid carcinoma comprises 55% to 75% of thyroid cancer and is the most common variety (295,363). Slow growing, this tumor may eventually spread to the paralaryngeal muscles, cervical lymph nodes, and lungs. Prognosis generally is favorable for patients of 40 years and younger but is decidedly poorer for patients older than 40 years. Almost 50% of patients have nodal metastases at the time of presentation; more than 20% who present with adenopathy have occult disease in the thyroid gland (363). The appearance of metastatic lymphadenopathy from papillary carcinoma may be variable and sometimes bizarre on imaging. Calcifications, cysts, or hyperdensity from hemorrhage or thyroglobulin may all be manifested (Fig. 4-75) (356,363). So-called water bag nodes, in which predominantly cysticappearing nodes resemble benign lesions, such as branchial cleft cysts, may be caused by papillary thyroid carcinoma. Tonsillar and nasopharyngeal carcinoma lymphadenopathy may also have this appearance. Thyroid carcinoma metastatic nodes are hypervascular, and metastases, especially within the brain, may spontaneously hemorrhage. Tracheal invasion, esophageal invasion, and carotid encasement worsen prognosis and should be evaluated when scanning these patients.

187

Figure 4-75 Papillary carcinoma of the thyroid. Computed tomography reveals an enhancing thyroid mass (arrowheads) extending into the left neck. A central hypodense region is noted. A tissue plane separates tumor from trachea (t). e, esophagus.

Follicular carcinoma represents 15% to 20% of thyroid malignancies. This cancer more commonly spreads hematogenously rather than through the lymphatics. Cystic degeneration is less common than in papillary carcinoma (451,458). Medullary carcinoma is rare and accounts for 5% of thyroid cancers (Figs. 4-76 and 4-77). These cancers involve the parafollicular or C cells of the thyroid that are responsible for calcitonin hormone production. Associated lymph node metastases are common, and bone metastases are more likely than in the other forms of well-differentiated thyroid malignancy (458). Medullary carcinoma can be associated with multiple endocrine neoplasia IIa (Sipple syndrome), with accompanying hyperplasia of the parathyroid glands and pheochromocytoma, as well as with multiple endocrine neoplasia IIb with mucosal neuromas and pheochromocytoma (458).

Figure 4-76 Medullary carcinoma of the thyroid gland. A large anterior neck soft tissue neck mass replaces the entire normal thyroid gland on computed tomography. The trachea (asterisk) is displaced to the right. Small flecks of calcium (arrowhead ) are deposited throughout the mass.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 188

188

Chapter 4

Figure 4-77 Medullary thyroid carcinoma. Well-defined, partially enhancing, right paratracheal mass (arrowheads) is seen on the enhanced computed tomography scan. Trachea (asterisk) is displaced to the left. The lesion abuts the right common carotid artery (arrow).

Anaplastic carcinoma of the thyroid gland accounts for 5% of thyroid malignancies and occurs most frequently in the elderly (391). Many patients have a long history of goiter prior to diagnosis. Symptoms of neck pain, hoarseness, dyspnea, and dysphagia may accompany a rapidly evolving neck mass (391). On CT images, patients usually demonstrate a large isodense mass with calcification (50%) and necrosis (75%). Almost 50% have necrotic lymphadenopathy and invasion of neighboring structures (391). Cervical lymphadenopathy is the presenting finding in many patients (366). Lymphoma of the thyroid may be primary or secondary. Primary lymphoma composes 8% of thyroid malignancies

and is usually seen in middle-aged and elderly patients. Women are much more frequently affected (219). Hoarseness and dysphagia occur in 25% of patients and recurrent laryngeal nerve paralysis in over 15%. More than 80% of patients have associated Hashimoto thyroiditis (310). Systemic lymphoma involves the thyroid gland in 10% of cases (310). Unfortunately, cervical MRI cannot reliably differentiate lymphoma of the thyroid from Hashimoto thyroiditis (281,393). On imaging, lymphoma presents as a large, homogeneous mass (Fig. 4-78). Neighboring structures may be infiltrated but usually with less frequency than anaplastic thyroid carcinoma (391). Although ultrasound and radionuclide scintigraphy are the primary imaging modalities in assessing thyroid masses (387), CT and MRI are useful for confirming the location of the mass within the gland, establishing the presence of intrathoracic extension, evaluating nodal disease, and assessing the airway (300). Surface coils are essential for MR examinations (122,274). Differentiation between benign and malignant primary thyroid masses is impossible on imaging, although associated lymphadenopathy, vocal cord paralysis, and bone or cartilage invasion obviously suggest malignancy (300) (Figs. 4-79 to 4-81). MRI signs suggesting tracheal invasion include an intraluminal mass and tumor abutting the trachea for greater than 180 degrees of its circumference (427). Apparent soft tissue signal within the tracheal cartilage may actually reflect tumor extension between the cartilage rings (427). The presence of calcification is not helpful and can be seen in about 30% of thyroid masses, regardless of etiology (304). MRI is useful in differentiating scar from residual or recurrent cancer in surgically treated patients (16,38). Tumor tends to be hypointense to isointense on T1-weighted images and isointense to hyperintense on T2-weighted images,

A

B Figure 4-78 Thyroid lymphoma. A, B: Proton density-weighted magnetic resonance images demonstrate an extensive tumor infiltrating the left and right neck. Both common carotid arteries (large arrows) are displaced posterolaterally. The left carotid is encased by tumor. The left internal jugular vein is not visualized and is most likely occluded. The posterior wall of the trachea (T ) is infiltrated with tumor. The cricoid cartilage (small arrows) is well visualized because of the high signal from medullary fat; right jugular vein (J); esophagus (e); sternocleidomastoid muscle (SCM); enlarged lymph node (arrowhead).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 189

Neck

Figure 4-79 Thyroid carcinoma. Postcontrast computed tomography image shows a large, irregular, low-density mass (M) destroying the left thyroid lamina and invading the left true vocal cord (arrowheads). More caudal images showed the mass arising from the left lobe of the thyroid.

whereas scar is hypointense on both T1-weighted and T2weighted images (280).

Parathyroid The parathyroid glands consist of two superior glands and two inferior glands. The superior glands are located on either side of the midline posterior to the upper lobes of the thyroid gland and usually between the inferior and superior thyroid arteries (295). The superior glands are derived from the fourth pharyngeal pouch (258). The inferior glands are in close proximity to the inferior poles of the thyroid and are derived from the third pharyngeal pouch (86,295). The glands may arrest anywhere along the path of embryologic migration or continue to migrate into the anterior mediastinum with the thymus and become ectopic

189

(86,387). Moreover, almost 25% of patients have supernumerary glands (86,258,456). Hyperparathyroidism results from excess parathyroid hormone and may manifest as hypercalcemia, osteopenia, renal stones, or psychiatric disturbance (258,456). Primary hyperparathyroidism may result from a single hyperfunctioning adenoma (80% to 85%), multiglandular hyperplasia (15%), hyperfunctioning parathyroid carcinoma (1% to 3%), or, rarely, from multiple adenomas in association with multiple endocrine neoplasia syndromes types I or IIa (295,387). Secondary hyperparathyroidism involves glandular hyperplasia in response to chronic hypocalcemia, such as in chronic renal insufficiency. Tertiary hyperparathyroidism involves autonomous multiglandular hyperfunctioning when the parathyroid no longer responds to the previous causative metabolic disorder (295). Parathyroid adenomas may occur anywhere along the path of embryologic migration from the upper neck to the mediastinum (300). Usually the adenoma is found in the tracheoesophageal groove and should not be confused with an ectatic vertebral artery on imaging (Fig. 4-82) (300,387). Almost 20% of glands, however, are ectopic, and more than 10% are found in the mediastinum (Fig. 4-83) (105). Successful surgical exploration for primary hyperparathyroidism exceeds 90% and rarely needs preoperative localization with imaging (105,300). Localization may be useful in patients with mild hypercalcemia, previous neck surgery or associated malignancy, obese necks, or associated high surgical risk medical conditions. Some authors contend that preoperative imaging decreases surgical time (38). Ultrasound and thallium-technetium subtraction scanning are usually preferred, although CT and MRI may be helpful in especially difficult cases (331,430). Undescended parathyroids may remain near the carotid bifurcation, around the hyoid bone, and imaging should be initiated cephalad at this level (86).

A

B Figure 4-80 Thyroid carcinoma. A: Enhanced computed tomography image demonstrates a large mass (M) infiltrating the right side of the neck and involving the right recurrent laryngeal nerve, resulting in right true vocal cord paralysis (white arrowheads). B: Similar findings are seen on T1-weighted magnetic resonance image; common carotid artery (black arrowhead); internal jugular vein (arrow); sternocleidomastoid muscle (SCM).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 190

190

Chapter 4

Figure 4-83 Parathyroid adenoma in superior mediastinum. An Figure 4-81 Cystic metastasis from thyroid carcinoma. A multiloculated, inhomogeneous low-density mass (arrows) is seen posterior to the left internal jugular vein (J) and sternocleidomastoid muscle (SCM); common carotid artery (C); clinically unsuspected thyroid carcinoma (arrowheads). (Courtesy of Dr. Jagan Ailnani, Carbondale, IL.)

On CT, the adenoma may lie posterior to the superior or inferior pole of the thyroid gland and may obliterate the retrothyroid-paratracheal-esophageal space (see Fig. 4-82) (121). About 25% of adenomas enhance (375). Detection largely depends on adenoma size (38). Interpretive pitfalls include tortuous neighboring arteries mimicking an adenoma, concomitant thyroid disease, calcified adenoma, superimposed hemorrhage, and streak artifacts from the shoulders (38). MRI is more sensitive than CT in detecting adenomas, especially in patients previously operated on with persistent hypercalcemia (105,289). T2-weighted images reveal parathyroid adenomas as hyperintense foci (Fig. 4-84) (195,200,373). The pitfalls of MRI are similar to those of CT and include misinterpretation of enlarged lymph nodes (243). MRI also may detect incidental intrathyroidal nodules, which can be confused with parathyroid lesions (113). Moreover, MRI cannot be used to distinguish adenomas from hyperplasia or carcinomas

Figure 4-82 Parathyroid adenoma. Adenoma (arrow) is seen in right tracheoesophageal groove; common carotid artery (c); internal jugular vein ( j ) ; esophagus (e); right lobe of thyroid gland (Th); trachea (T ) .

elliptical mass (arrows) is seen adjacent to the left side of the trachea (T ) ; esophagus (e); common carotid artery (c); internal jugular vein ( j ); subclavian artery (s); vertebral artery (v).

(373,387). Elevated parathyroid hormone levels in needle aspirates of cervical masses in patients with hyperparathyroidism confirm the presence of a parathyroid adenoma (85,315).

RETROPHARYNGEAL SPACE The retropharyngeal space lies posterior to the visceral space and is sandwiched between the middle and deep layers of the deep cervical fascia (see Fig. 4-7). The retropharyngeal space extends from the base of the skull to the mediastinum and serves as a potential conduit for spread of neck pathology into the chest (73,74). The retropharyngeal space is divided into suprahyoid and infrahyoid compartments (73). The suprahyoid compartment contains lymph nodes and fat, whereas the infrahyoid compartment only contains fat (73,74). The lymph nodes of the suprahyoid component of the retropharyngeal space form paired medial and lateral chains. The lateral chain is also known as the nodes of Rouvier (73). Therefore, retropharyngeal lymphadenopathy only occurs above the hyoid and tends to remain unilateral or bilateral, sparing the midline (Fig. 4-85). In contradistinction, infections and direct invasion of cancer may involve both the suprahyoid and infrahyoid portions and the midline (73) (Fig. 4-86). A “danger space” posterior to the retropharyngeal space has been described (152). The danger space cannot be reliably differentiated from the retropharyngeal space on imaging and is therefore combined with the retropharyngeal space for discussion. Retropharyngeal masses lie anteriorly to the perivertebral space, posteromedially to the parapharyngeal space, and medially to the carotid arteries (74,152). The perivertebral muscles may be compressed and laterally splayed (152). Common retropharyngeal lesions include inflammatory lymphadenopathy and abscesses (434). Differentiation between retropharyngeal abscess and adenitis is difficult on CT because both processes cause hypodense regions within the inflammatory mass. Ultrasound has

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 191

Neck

A

191

B Figure 4-84 Parathyroid adenoma. A: On T1-weighted magnetic resonance image, a small mass (white arrow) is present posterior to the left lobe of the thyroid (Th). It is similar in signal intensity to the thyroid gland; common carotid artery (C ) ; internal jugular vein (J ) . B: On a T2-weighted image, the adenoma (arrow) is hyperintense and easily distinguishable from the adjacent thyroid gland (Th). (From Spritzer CE, Gefter WB, Hamilton R, et al. Abnormal parathyroid glands: high-resolution MR imaging. Radiology l987;l62:487–49l, with permission.)

been advocated for further evaluation (127). Adenitis is usually managed with antibiotics. Abscesses developing from the rupture of nodes into the retropharyngeal space or adjacent spaces are surgically managed (152). Identification of a ring enhancing fluid collection remains the best predictor for drainable fluid, although surgical correlation is disappointing in many cases. Suppurative collections are often accompanied by extensive retropharyngeal edema. The latter should not be considered drainable. Neoplastic lymphadenopathy, especially from advanced maxillary sinus, oropharyngeal, and hypopharyngeal carcinoma, and direct extension from mucosal carcinoma, such as nasopharyngeal cancer, can involve the retropharyngeal space (159,429). Hemangioma, hematoma, and lipoma also occur in this space (Fig. 4-87) (152). Retropharyngeal soft tissue emphysema may be caused by direct trauma,

infection, foreign body ingestion, or assisted ventilation (74). Tortuous carotid arteries can occasionally meander medially into the retropharyngeal space and should not be clinically or radiographically mistaken for masses (see Fig. 4-20) (336,338). Fluid from lymphatic or venous obstruction can mimic an abscess (Fig. 4-88) (152).

POSTERIOR CERVICAL SPACE The posterior cervical space abuts the carotid space posterolaterally and is sandwiched by the sternocleidomastoid muscle anterolaterally and the paraspinal muscles posteromedially (see Fig. 4-7B and C). Fascial boundaries include the deep layer of the deep cervical fascia medially and the superficial layer of the deep cervical fascia posterolaterally

Figure 4-86 Retropharyngeal tuberculous abscess. Computed Figure 4-85 Retropharyngeal lymph nodes. Enhanced computed tomography in a patient with nasopharyngeal carcinoma demonstrates an enlarged left retropharyngeal lymph node (arrowheads). Biopsy revealed metastatic carcinoma.

tomography demonstrates a hypodense fluid collection involving the retropharyngeal space (asterisks). The midline is involved. The airway is displaced anteriorly. The perivertebral muscles (arrows) are flattened posteriorly. Surgical aspiration and cultures revealed tuberculosis.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 192

192

Chapter 4

Figure 4-89 Lipoma, posterior cervical space. Enhanced com-

Figure 4-87 Hemangioma. Enhanced computed tomography demonstrates a large expansive densely enhancing homogeneous retropharyngeal mass (asterisk). The airway is displaced anteriorly. The perivertebral muscles (m) are flattened posteriorly.

(147,354,438). The posterior cervical space corresponds to the occipital and subclavian divisions of the posterior cervical triangle (287). The primary components of this space are fat, the spinal accessory and dorsal scapular nerves, and the spinal accessory lymph nodes of the deep cervical chain.

Figure 4-88 Lymphocele, retropharyngeal space. On the enhanced computed tomography image, a nonenhancing fluid density (black arrows) symmetrically expands the suprahyoid retropharyngeal space. This patient with sickle cell anemia had a chronic left subclavian catheter with angiographically demonstrated central venous thrombosis involving both innominate veins and the proximal left subclavian vein. The fluid was chylous. Note the expansion of the hematopoietically active mandible (open arrows) and the prominent submandibular ducts (arrowheads) coursing through the sublingual spaces. Submandibular salivary gland (asterisks).

puted tomography scan reveals a lipid density mass (arrowheads) in the posterior cervical space. The right sternocleidomastoid muscle (m) is lifted anterolaterally and the paraspinous muscles (M) are compressed posteromedially. The posterior cervical space corresponds to the posterior triangle of the neck and is a common location for lipomas involving the neck.

Lesions originating within the posterior cervical space are embedded within the fat of this space and displace the carotid space anteromedially and the perivertebral space posteromedially (287). Typical lesions arising in this space include spinal accessory chain lymphadenopathy from metastatic squamous carcinoma and lymphoma, lipomas, liposarcomas, cystic hygromas, and branchial cleft cysts (Figs. 4-89 and 4-90). Occasionally, hypertrophy of the levator scapulae muscle may mimic tumor (287).

Figure 4-90 Cystic hygroma, posterior cervical space. A large water attenuation mass (asterisk) occupies the posterior cervical space on enhanced computed tomography. The left internal jugular vein (arrow) is displaced anteromedially and the sternocleidomastoid muscle (m) is displaced laterally. The muscles (M ) of the posterior compartment of the perivertebral space are displaced posteromedially.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 193

Neck

PERIVERTEBRAL SPACE The perivertebral space is formed by the deep cervical fascia. Fascia attaches to the transverse processes of the cervical vertebrae, dividing this space into anterior and posterior compartments (see Fig. 4-7) (36,74,438). The anterior compartment contains the vertebral bodies and spinal cord, vertebral arteries, phrenic nerve, and perivertebral and scalene muscles. The posterior compartment contains the posterior vertebral elements and paraspinous muscles (74). Many sources prefer to use the term perivertebral space for this compartment because it aptly describes the full extent of its location (151,438); however, the older term, prevertebral space, is still in use by some investigators. Perivertebral space lesions usually arise in the vertebral body, intervertebral disc spaces, or perivertebral or paraspinous muscles (151,438). Examples include vertebral osteomyelitis, necrotizing fasciitis and metastases, and rarer lesions, such as chordoma and nerve sheath tumors (Fig. 4-91), especially plexiform neurofibromas (74). On imaging, perivertebral lesions anteriorly displace the retropharyngeal space and anterior border of the perivertebral muscles and posterolaterally displace the posterior triangle fat (151). In differentiating primary perivertebral muscle lesions, such as sarcomas and inflammatory masses, from bulky lymphadenopathy,

193

it is useful to recall that lymph nodes arise in the fat planes rather than the muscle beds themselves.

LYMPH NODES OF THE NECK Anatomy Cervical lymphatics are critical conduits that drain the skin and mucosal surfaces of the head and neck. Nearly 40% of all lymph nodes in the body are located above the clavicles; only the orbit and cervical muscles do not have direct lymphatic connections (357). In health, lymphatics maintain balance between intravascular and extravascular fluid. In disease, the lymphatics are a pathway for the spread of inflammation and neoplasms (284). Nodes are usually embedded in the fat planes that surround vessels and separate major cervical muscles. All cervical lymph drainage passes through at least one lymph node on the journey to the venous system. Ten major groups of cervical lymph nodes are recognized (183,231,303,313,438). These groups are named for the structures in proximity to the nodes. These nodes form a sentinel ring around the base of the skull that is drained by four chains. Conceptually, this arrangement resembles a tabletop supported by four legs

B

A Figure 4-91 Neurofibroma in a patient with neurofibromatosis 1. A large hypointense soft tissue mass nearly fills the central vertebral canal at C-2 on the T1-weighted magnetic resonance image (A). The lesion intensely enhances following administration of intravenous contrast (B). The mass extends through and enlarges the left C1-2 neural foramen (white arrows). It grows into the anterior and posterior compartments of the perivertebral space and encases the left vertebral artery (large arrowheads). The retropharyngeal space (small black arrows) is displaced anteromedially, and the parapharyngeal space (large black arrows) is displaced anterolaterally. The lesion wraps around the C-2 vertebral body and therefore is anterior to the perivertebral muscles (asterisks). Lesions arising in the anterior compartment of the perivertebral space displace the perivertebral muscles anteriorly. The paraspinous muscles are displaced away from the spine (small arrowheads). (From Brown J, Wippold FJ 2nd. Practical MRI: a teaching file. New York: Lippincott-Raven Press, 1996:200–220, with permission.)

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 194

194

Chapter 4

Figure 4-92 Schematic conceptualization of the cervical lymph nodes. The outer ring forms the table surface and represents the sentinel chains at the base of the skull including the occipital, mastoid, parotid, submandibular, facial, submental, and sublingual nodal groups. The shaded inner C-shaped structure represents the deep retropharyngeal nodes that extend to the hyoid bone. All of these groups drain into the paired anterior and lateral chains depicted as the legs of the table.

(Fig. 4-92). The table surface is formed by the first six groups (occipital, mastoid, parotid, submandibular, facial, and submental), which are superficial and easily palpable; the sublingual nodes (group 7), which are deep in the floor of the mouth; and the deep retropharyngeal nodes (group 8). The paired anterior (group 9) and lateral (group 10) cervical chains of lymph nodes drain the sentinel ring and form the legs of the table. Of these groups, the submental, submandibular, retropharyngeal, and lateral cervical chains play especially important roles in the spread of head and neck disease (Fig. 4-93). The submental group is located inferior to the anterior mandible and mylohyoid muscle and between the digastric muscles (Figs. 4-94 and 4-95). These nodes drain the chin, lower lip, cheeks, and floor of the mouth. Lymph then flows toward the submandibular and lateral cervical chain. The submandibular group is located in the submandibular space and drains the lateral chin, lips, cheeks, nose, anterior tongue, submandibular salivary glands, floor of the mouth, and submental nodes (see Fig. 4-14). Flow subsequently drains into the lateral cervical chain. The retropharyngeal nodes are located in the suprahyoid retropharyngeal space, along the lateral borders of the longus capitus muscle, and are responsible for draining the nasal cavity, paranasal sinuses, nasopharynx, and deep mucosa (Figs. 4-85 and 4-96) (233,234). The occipital, mastoid, and facial nodes drain the skin of the scalp, ear, and face and empty into the anterior and lateral cervical chains. The sublingual nodes are found in the sublingual space and drain the tongue and floor of the mouth. A lateral group follows the course of the lingual artery, and a median

Figure 4-93 Diagram of the cervical lymph nodes. Notable groups of the neck include: occipital (1), mastoid (2), parotid (3), submandibular (4), facial (5), submental (6), sublingual (7), retropharyngeal (8), anterior cervical (9), and posterior cervical (10). The sublingual and retropharyngeal nodes are concealed in the deep structures of the head and neck.

group lies between the genioglossus muscles (Fig. 4-97) (303). Flow is toward the submental, submandibular, and lateral cervical chains. The parotid nodes are actually embedded within the parotid gland and usually are not seen unless enlarged (Fig. 4-96). These nodes serve the scalp,

Figure 4-94 Submental lymph node. Enhanced computed tomography in a patient with right radical neck dissection (white arrowheads) reveals a submental lymph node (asterisk) embedded in the fat between the digastric muscles (white arrows).

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 195

Neck

195

Figure 4-97 Sublingual lymph nodes. Computed tomography in a patient with lymphoma and extensive cervical lymphadenopathy demonstrates a small collection of lymph nodes (arrows) in the sublingual space adjacent to the sublingual salivary glands and submandibular duct and in the median low density space between the genioglossus muscles.

Figure 4-95 Submental lymph node. Computed tomography demonstrates a submental lymph node (asterisk) imbedded in the fat of the submental triangle of the neck.

face, parotid gland, and buccal mucous membrane and drain into the lateral cervical chain. The lateral cervical chain is subdivided into the superficial and deep lateral cervical nodes. The superficial group follows the course of the external jugular vein, is easily palpable, and therefore is not usually examined with imaging. The important deep group is further divided into the spinal accessory, transverse cervical, and internal jugular groups. The spinal accessory nodes are found within the fat of the posterior cervical triangle and posterior cervical space lateral and posterior to the spinal accessory nerve between the trapezius and the sternocleidomastoid muscles. The transverse cervical group is seen in the supraclavicular region.

The internal jugular group is deep to the sternocleidomastoid muscle and follows the course of the internal jugular vein. High internal jugular nodes extend from the base of the skull to the carotid bifurcation. Because the carotid bifurcation is variable and occasionally difficult to identify on scans, the hyoid bone can be used as a boundary. The jugulodigastric node is a prominent node in this region and may be 1.5 cm in diameter (Fig. 4-98). The middle jugular nodes extend from the carotid bifurcation to the omohyoid muscle. Again, radiologists can use the cricoid cartilage as a more consistent landmark. Finally,

Figure 4-98 Internal jugular chain lymph node. Computed toFigure 4-96 Computed tomography in a patient with lymphoma reveals a necrotic left retropharyngeal lymph node (arrowhead). An enhancing intraparotid lymph node (black arrow) is also evident.

mography in patient with tonsillar carcinoma shows an enlarged right internal jugular lymph node (asterisk). A small left internal jugular lymph node (arrow) also is noted.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 196

196

Chapter 4

all nodes in the internal jugular group from the omohyoid muscle to the clavicle. They lie anterior to the posterior margin of the sternocleidomastoid muscle and lateral to the carotid arteries. Level V consists of spinal accessory and transverse cervical nodes that occupy the posterior cervical triangle. They lie posterior to the posterior margin of the sternocleidomastoid muscle. Level VA extends from the skull base to the inferior margin of the cricoid arch, and level VB extends from the cricoid arch to the clavicle. Level VI contains the pretracheal, prelaryngeal, and paratracheal nodes. These nodes lie between the carotid arteries from the hyoid bone to the manubrium. Finally, level VII includes the nodes in the tracheoesophageal groove and upper mediastinum. They lie between the carotid arteries below the manubrium and above the innominate vein (365).

Pathology

Figure 4-99 Lymph node levels based on the American Joint Committee on Cancer guidelines. Level I corresponds to the submental and submandibular nodes. Level II refers to the high jugular nodes of the deep cervical chain. Level III corresponds to the middle internal jugular nodes of the deep cervical chain. Level IV includes the lower deep cervical chain nodes. Level V includes the spinal accessory nodes and the transverse cervical chain nodes. Level VI contains the pretracheal and paratracheal nodes. Level VII includes the nodes in the tracheoesophageal groove and upper mediastinum.

the low jugular nodes span from the omohyoid muscle to the clavicle. The nodes of Virchow are the most inferior nodes in the deep cervical chain (357). The American Joint Committee on Cancer (AJCC) and the American Academy of Otolaryngology–Head and Neck Surgery have established guidelines using terminology that divides the lymph node groups into a series of levels that have prognostic importance (Fig. 4-99) (137,153,189,357). Recently, some modifications based on imaging have also been proposed (365). Level I consists of the submental and submandibular nodes. These nodes lie above the hyoid bone, below the mylohyoid muscle, and anterior to the posterior margin of the submandibular gland. Level IA nodes are the submental nodes and level IB are the submandibular nodes. Level II includes the internal jugular chain nodes extending from the base of the skull to the carotid bifurcation (hyoid bone). They lie posterior to the submandibular gland and anterior to the posterior margin of the sternocleidomastoid muscle. Level IIA nodes are anterior, medial, or lateral to the internal jugular vein. If posterior to the vein, the nodes are inseparable from the vein. Level IIB nodes are posterior to the internal jugular vein, with a fat plane that separates the vein from the nodes. Level III corresponds to the internal jugular nodes from the carotid bifurcation to the omohyoid muscle (cricoid cartilage). They lie anterior to the posterior margin of the sternocleidomastoid muscle. Level IV refers to

The detection of lymphadenopathy in head and neck cancer patients is critically important because the presence of nodal metastases is associated with a 50% reduction in 5-year survival (224,356). More than 15% of clinically N0 patients may have histologic metastases on pathologic examination (184,379). Imaging is indispensable in the evaluation of cervical lymphadenopathy (49,110,226,236,247,380). Between 20% and 30% more metastases are noted on imaging than are originally detected by palpation (81). In 5% of patients with clinically N0 necks, CT will upstage to N1 (234). Moreover, extracapsular spread, carotid invasion, and conglomerate nodes may not be fully appreciated on manual examination. The retropharyngeal nodal group is especially difficult to evaluate clinically. The role of the radiologist is to recognize cervical lymphadenopathy and associated findings such as extracapsular involvement, to anticipate nodal chains at risk given a primary tumor site, and to suggest a primary tumor site in the patient who presents with clinically isolated lymphadenopathy and an occult primary tumor. Imaging criteria for lymphadenopathy are based on nodal size, internal heterogeneity, presence of clusters, shape, and associated findings (81,189,234,236,356,357,410). Nodes in levels I and II generally are larger than nodes in lower levels. The maximum transaxial diameter of a node in levels I or II should not exceed 1.5 cm, and the minimum transaxial diameter should not exceed 1.1 cm. For levels III to VII, maximum and minimum transaxial diameter should be no greater than 1.0 cm (52,414). On the basis of these criteria, 80% of enlarged nodes in a patient being staged for cancer will be metastatic and 20% will be hyperplastic (224). Inflammatory conditions also may enlarge nodes. Internal lymph node heterogeneity is one of the most reliable criteria for recognizing lymphadenopathy (224). Normal cervical nodes appear homogeneous and of soft tissue attenuation or intensity on imaging. Central regions within nodes displaying hypodensity on CT, hypointensity on T1-weighted MR images, and hyperintensity on T2-

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 197

Neck

Figure 4-100 Necrotic lymphadenopathy. Computed tomography in a patient with transglottic carcinoma of the larynx demonstrates large bilateral necrotic internal jugular lymph nodes (arrows) with central hypodensity.

weighted MR images should be regarded as abnormal and usually signify necrosis (Fig. 4-100). Necrosis generally is proportional to nodal size; however, this finding should be considered abnormal regardless of nodal size (109). Metastasis and suppuration remain the most common causes of necrotic lymphadenopathy. Metastatic nodes tend to develop firm and nontender masses over a period of weeks to months, whereas abscessed nodes rapidly evolve into exquisitely painful swellings. An exception to the guideline of internal heterogeneity, as a marker for malignancy or acute infection, is the observation that fat may occupy the hila of normal nodes and may replace portions of abnormal nodes previously inflamed or irradiated (Fig. 4-101). Additionally, the occasionally encountered benign cystic lesion, such as an inflamed branchial cleft cyst, may be difficult to differentiate from necrotic lymphadenopathy (367). Similarly, nodal homogeneity does not exclude metastases because homogeneous nodes may harbor nonnecrotic tumor cells. Nodes involved with lymphoma are

Figure 4-101 Fat in lymph node hilum. Computed tomography scan showing a normal ovoid posterior submandibular lymph node (black arrow) with an ovoid collection of fat (white arrowheads) within its hilum.

197

Figure 4-102 Lymphoma with lymph node clusters. Computed tomography in a patient with lymphoma showing numerous round lymph nodes (arrowheads) clustered in the lateral cervical chain. Although many of these individual nodes are normal in size, the large number of nodes aggregated within a nodal chain is abnormal. Extensive lymphadenopathy is present bilaterally.

typically homogeneous (314). Enhanced CT is better than unenhanced MRI in establishing nodal necrosis (459); however, use of intravenously administered gadolinium agents can augment the effectiveness of MRI (181,413). Clusters are defined as three or more contiguous illdefined nodes within the same level ranging from 8 mm to 15 mm in size. Clusters may be seen in inflammation, cancer, or lymphoma (Fig. 4-102). Small cancerous nodes, seemingly normal by size criteria, may be clustered with larger obviously malignant nodes (109,236). Shape is no longer thought to be reliable in differentiating normal from pathologic nodes. Round nodes tend to be neoplastic whereas elliptical or bean-shaped nodes are generally normal or hyperplastic; however, many exceptions may be encountered. Some authors have suggested that the ratio of maximal longitudinal to maximal axial diameter of enlarged nodes, as calculated with spiral CT, may be used to differentiate malignant from reactive nodes. Nodes with ratios less than two and nearly spherical in shape are more likely to be malignant than nodes with ratios of two or more, which are more likely benign (Fig. 4-103) (378). The presence of extracapsular spread of disease implies a 50% reduction in 2-year patient survival compared with the survival of patients with metastases confined to the nodes (184). Tumor spread beyond the capsule of a node is manifested by capsular enhancement, ill-defined nodal margins, obliterated fat planes surrounding the node, and edema or thickening in the adjacent soft tissues (Fig. 4-104) (300,356). Likelihood of extracapsular spread generally increases with nodal size. More than 50% of pathologic nodes 2 to 3 cm in diameter and almost 75% of nodes greater than 3 cm in diameter have microscopic evidence of extracapsular involvement. The carotid artery should be carefully examined in the presence of internal jugular node extracapsular invasion because tumor can abut and potentially involve this artery, thus altering clinical management.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 198

198

Chapter 4

Figure 4-103 Round lymph nodes. Enhanced computed tomography in a patient with lymphoma demonstrates clusters of round bilateral submandibular lymph nodes (black arrows). Nearly spherical lymph nodes also are identified in the internal jugular chain (black arrowheads) and external jugular chain (white arrowheads).

Mimics of extracapsular nodal disease include the changes observed in radiated or surgically altered nonmalignant nodes (224). No clear consensus has emerged on the superiority of CT or MRI in the assessment of extracapsular spread (49,459). Additional imaging signs of lymphadenopathy are the presence of enhancement and calcification within nodes. Enhancement on MR or CT images is abnormal. Calcification of a node may be seen in granulomatous disease such as tuberculosis, previously radiated neoplastic nodes, and metastatic thyroid carcinoma (357,363). Although controversial, CT generally is preferred to MRI in the evaluation of cervical lymphadenopathy (224,459). MRI probably is superior in detecting retropharyngeal nodes (392). The CT criteria for lymphadenopathy are well established. On MR imaging, the appearances of

Figure 4-104 Extracapsular lymphadenopathy. Enhanced computed tomography shows a large, necrotic right internal jugular lymph node mass (asterisk). The enhanced margins are irregular (black arrowheads) and the skin is infiltrated (white arrows). The primary lesion (black arrows) originated in the base of the tongue and invaded the right tonsil.

necrotic and nonnecrotic metastatic and reactive nodes may be similar (356,392). The addition of fat-suppression techniques and enhancement increases confidence in the detection and analysis of cervical nodes by MRI (181,356). Investigations are currently exploring the use of dextrancoated superparamagnetic iron oxide as an intravenous MR contrast agent for differentiation of benign from malignant nodes. Normal functioning nodes trap the iron oxide and become hypointense, whereas malignant nodes remain isointense (8,10). The AJCC has established criteria for nodal staging to be used in conjunction with the TNM classification. In general, progressively higher stages imply larger and more numerous abnormal nodes. The risk of nodal metastases is site specific and generally associated with the richness of the draining lymphatic capillary bed (184). Certain regions, such as the lip, glottis, and paranasal sinuses, have relatively low risk for associated lymphadenopathy, and other regions, such as the pyriform sinus and supraglottis, have high risk (249). Midline tumors are at greater risk for bilateral lymph node metastases. Lymphatic drainage typically proceeds in an orderly craniocaudal direction, and therefore drainage sites distal to the primary lesion should be evaluated for metastases. Exceptions to this rule are in the previously radiated or surgically altered neck in which collateral channels may lead to metastatic deposits in seemingly atypical chains (183). Tongue cancers are richly supplied by lymphatics and have a high frequency of bilateral lymph node metastases. The submandibular, high internal jugular, and middle internal jugular nodes (levels I to III) are at risk in tongue lesions. Interestingly, the further anterior in the oral tongue the primary tumor, the more caudally the likely involved cervical nodes (284). Floor of the mouth lesions usually involve the submandibular and high internal jugular nodes. Soft palate, tonsillar, and supraglottic cancers typically involve the high and middle internal jugular chains. As tumors become more advanced, additional drainage chains are jeopardized.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 199

Neck

Occult malignancies presenting with cervical lymphadenopathy pose an especially difficult management problem. Causes for the inability to clinically locate the primary lesion include small size at presentation, inaccessibility to clinical inspection, predominately deep submucosal pattern of spread, and, conceivably, even spontaneous regression of the original pathology (81,154). Imaging is useful in identifying the location of the occult primary neoplasm, suggesting a possible biopsy site, and assessing the extent of the disease (81). In patients with metastatic nodes originating from an unknown upper aerodigestive tract site and a negative blind biopsy, CT can successfully suggest the primary tumor in almost 25% of cases (81,269). Patients with occult squamous cell carcinoma metastatic to level I or level II nodes should undergo imaging of the nasopharynx, oropharynx, and hypopharynx (227,231). Of occult primary tumors initially presenting with lymphadenopathy, the nasopharynx is the most commonly encountered site (357). In approaching these patients, special attention should be given to subtle asymmetries in the oral, pharyngeal, and laryngeal mucosa on CT and MR images so as to direct biopsies of otherwise clinically undetectable abnormalities. Submandibular lymphadenopathy suggests the tongue and floor of the mouth as primary sites. Upper cervical node involvement should also prompt inspection of the base of the tongue and tonsillar region. Middle and lower internal jugular node involvement suggests larynx, hypopharynx, or thyroid primary sites. Midline and paratracheal nodes may be involved in thyroid, larynx, and even lung cancers. Involvement of the supraclavicular nodes may be the result of spread of thoracic and even abdominal malignancies. In the past decade, 2-[fluorine-18] fluoro-2-deoxy-D-glucose (FDG) single-photon emission computed tomography (SPECT) has yielded promising results as an adjunct to conventional cross-sectional imaging in evaluating the occult malignancy (258). Lymphoma is the second most common malignancy in the neck and is the most common extralaryngeal malignancy. Hodgkin lymphoma is one of the most common neck malignancies in children (300,304). Lymphoma of the neck commonly involves the adenoids and lingual and pharyngeal tonsils (Waldeyer’s ring). The cervical lymph nodes, especially the internal jugular and spinal accessory nodes, also usually are involved. Extranodal sites such as the thyroid gland also may harbor malignant disease. On imaging, clusters of round, nonnecrotic nodes are often observed (Fig. 4-105) (314). Large cell lymphoma may cause especially large nonnecrotic lymph nodes (Fig. 4-106) (314). Attenuation and enhancement properties are variable (356). Necrosis may be seen following therapy. Almost 50% of AIDS patients have neck abnormalities (111). Homogeneous cervical lymphadenopathy in a young patient with associated nasopharyngeal lymphoid hyperplasia and lymphoepithelial parotid lesions should prompt consideration of HIV infection (seee Fig. 4-18) (59,198). Of AIDS patients with nonHodgkin lymphoma of the

199

Figure 4-105 Lymphoma. Multiple, enlarged, non-necrotic lymph nodes are seen; hyoid bone (H).

neck, most are severely immunodeficient, exhibit extranodal disease, have aggressive tumors, and respond poorly to treatment (48). Necrotic nodes in these patients should also prompt evaluation for a primary squamous cell carcinoma or Kaposi sarcoma (111). Tuberculous lymphadenitis typically affects young patients and may present as a conglomerate mass of necrotic, enhancing, and often calcified nodes associated with inflammatory changes in the skin and adjacent fascial planes (111,171,196,204). Unlike acute suppurative lymphadenitis, tuberculosis is usually painless and may even resemble aggressive cancer or lymphoma. Other conditions producing cervical lymphadenopathy include infectious mononucleosis, sarcoidosis, sinus histiocytosis with massive lymphadenopathy, and Castleman disease (171,356,411).

Figure 4-106 Lymphoma. An enlarged non-necrotic lymph node (arrows) is seen on computed tomography posterior to the right internal jugular vein (J). C, common carotid artery.

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 200

200

Chapter 4

CYSTIC NECK LESIONS Cystic neck lesions challenge the radiologist because the imaging features are often quite striking and offer multiple clues for diagnosis. Of the several cardinal features considered in the evaluation of the cystic neck mass, location is of prime importance. Many processes are associated with specific anatomic compartments, therefore narrowing the range of diagnostic possibilities. For example, ranulas always begin in the sublingual space by definition and are therefore easily included in the discussion of floor of the mouth processes and excluded from the differential consideration of diseases elsewhere. Most cystic lesions are infrahyoid in location. The clinical history also is important. A painful, rapidly evolving lesion in a young patient may well be an abscess, whereas the slowly growing, nontender mass in an elderly smoker is more likely lymphadenopathy. The age of the patient is helpful in sorting possible diagnoses. Cystic lesions in the first two decades of life tend to be benign and related to congenital or inflammatory conditions, in the third and fourth decades they tend to be malignant, and beyond the fourth decade they tend to be metastatic (68,304). The imaging density or intensity of the lesion may contribute information about its contents. Lipomas, which may masquerade as cystic, can be discerned by recognizing the imaging signature of fat. Hemorrhage also may influence signal intensity (386). Inspection of the surrounding tissue is also valuable. Edematous subcutaneous fat and obscured soft tissue planes may indicate infection or possibly an infiltrating neoplasm. Finally, multiplicity is an important finding in lesions such as metastases, but would not be expected in congenital conditions such as thyroglossal duct cysts. Lymphadenopathy is the most common cause of a cystic neck mass, considering all ages at presentation. Pathologic lymph nodes can present along any of the lymph node chains, although internal jugular and spinal accessory nodal groups are the most common (Fig. 4-107). These lesions tend to be hard, fixed, and painless. The cystic features are the result of necrosis, and large nodes adjacent to the anterior border of the sternocleidomastoid muscle may be confused with a branchial cleft cyst. Etiologies for the necrotic node may be carcinoma or inflammation; lymphoma rarely produces cystic change. Occasionally, the necrotic nodes of papillary thyroid carcinoma have smoothly regular walls (363). Abscesses are usually rapidly developing and painful and may be associated with fever and leukocytosis. Abscesses frequently involve the nodal groups draining the primary site of infection (Fig. 4-108). One third of cases of suppurative cervical lymphadenopathy are associated with pharyngitis and tonsillitis, one third with dental infection, and one third with dermatologic and otologic infection. Imaging distinguishes the ill-defined edema of cellulitis, which is best treated medically, from the coalesced and cystic abscess, which requires surgery. Abscesses frequently have

Figure 4-107 Metastatic lymphadenopathy. Enhanced computed tomography shows a large cystic appearing right internal jugular chain lymph node (asterisk) with a surrounding rim of enhancement. A second necrotic node (arrow) within the right submandibular space suggests the correct diagnosis of metastases rather than other causes of cystic neck masses.

Figure 4-108 Abscess. Enhanced computed tomography demonstrates confluent, necrotic submandibular lymph nodes (asterisk) draining the site of an infected dental procedure. Cultures grew Streptococcus. The subcutaneous fat is infiltrated (arrow) and the airway is encroached upon.

thick ring-configured enhancement and may be multiloculated. Untreated infection may dissect along fascial planes. Salivary lesions may be cystic. Warthin tumors are usually found in older men and may be bilateral, multifocal, and cystic. Pleomorphic adenomas characteristically present as solitary lesions and also may be cystic (Fig. 4-13). Mucusfilled retention cysts derived from obstructed or traumatized

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 201

Neck

201

Figure 4-110 Papillary carcinoma arising in thyroglossal duct cyst. A multilobed cystic mass is seen anterior to the supraglottic portion of the larynx. Focal areas of calcification (arrows) and thickened soft tissue septa (arrowheads) are seen within the mass; common carotid artery (c); internal jugular vein (J). (Courtesy of Dr. Edward Ragsdale, Alton, IL.)

Figure 4-109 Mucocele, right submandibular salivary gland. Enhanced computed tomography reveals a large mass of water density (asterisk) in the right suprahyoid neck; normal residual portion of the right submandibular salivary gland (s). At surgery, the mucocele extended eccentrically from the anterior border of a chronically inflamed right submandibular salivary gland.

salivary ducts may become quite large (Fig. 4-109) (436). Lymphoepithelial lesions of the parotid glands are seen in patients with AIDS and are frequently multicentric and bilateral (see Fig. 4-18) (150). Thyroglossal duct cysts comprise 40% of primary parenchymal neck masses and 70% of congenital neck abnormalities (302,350). Almost 50% are discovered prior to age 30 years. Thyroglossal duct cysts are derived from vestiges of the developing thyroid gland. Normally, thyroid tissue originates in the tongue base and migrates along a pathway known as the thyroglossal duct to the anterior trachea. Usually the duct involutes; however, aberrant cystic remnants may persist anywhere along the path of migration and present as a neck mass in a young adult (302,316). The cyst may be suprahyoid in location (20% to 25% of cases), at the level of the hyoid (15% to 50%), or infrahyoid in location (25% to 65%) (338). Surrounding inflammation from an upper respiratory tract infection may prompt the conversion of a thyroglossal remnant into a cyst (202,446). These cysts may be lined with either squamous cell or respiratory epithelium and may themselves become inflamed (25,386). Carcinomas within the cyst occur in less than 1% of patients; however, when present, 80% will be the papillary variety (Fig. 4-110) (25,202). An eccentric mass in the cyst wall or a solid mass in the expected course of the thyroglossal duct should prompt consideration of a complicating cancer, especially if calcified (128). Occasionally, thyroglossal duct cysts project

into the pre-epiglottic space and may be mistaken clinically for a submucosal mass or laryngocele (350). Almost 65% of thyroglossal duct cysts are infrahyoid, 20% are suprahyoid, 15% are found at the level of the hyoid, and 75% are midline (25,304,355,446). The laterally positioned cysts located at the level of the thyroid cartilage may appear to be embedded in the paralaryngeal strap muscles (Fig. 4-111) (367). This appearance may resemble a “snake swallowing an egg” (42) and can help differentiate a thyroglossal cyst from a second branchial cleft cyst, which is usually posterolateral to these muscles (Fig. 4-112). Additionally, the path of migration of the thyroid hooks anteriorly and then inferiorly and posteriorly to the hyoid bone, often leaving a notch or defect in the hyoid bone (355,367) (Figs. 4-113 and 4-114). Surgical removal of the body of the hyoid bone is recommended in these patients to avoid recurrence (245). A laryngeal mucocele is a fluid-filled laryngocele and can be seen with obstruction of the ostium of a laryngeal saccule. An enhancing wall usually indicates infection (Figs. 4-115 and 4-116). Approximately 15% may be associated with cancer. External or mixed laryngeal mucoceles protrude through the thyrohyoid membrane. The endolaryngeal origin of this lesion differentiates this lesion from a lateral thyroglossal duct cyst (355). Branchial cleft cysts usually are painless, fluctuant, lateral neck masses that arise from anomalous branchial

5063_Lee_Ch04pp0145-0224 10/13/05 12:06 PM Page 202

202

Chapter 4

Figure 4-111 Thyroglossal duct cyst. A cystic left mass (asterisk) embedded within the paralaryngeal strap muscles on T1-weighted magnetic resonance image. The fluid is mildly hyperintense; the subcutaneous fat is normal. (From Brown J, Wippold FJ 2nd. Practical MRI: a teaching file. New York: Lippincott-Raven Press, 1996: 200–220, with permission.)

arch development (155,202,259,316,338,349,355,367). These lesions usually present in young adults, and almost 10% can be bilateral (155,436). Branchial cleft cysts also may be associated with fistulas and may become intermittently infected. The second branchial cleft accounts for almost 90% of cases (355,446). The rare first branchial cleft cyst is located near the pinna and parotid gland. Third and fourth branchial cleft cysts are seldom encountered; the

former may present in the posterior cervical space (287,355,436). On imaging, branchial cleft cysts are thin walled, filled with serous or mucoid fluid, and typically do not enhance (Figs. 4-117 and 4-118). With infection, the wall enhances and thickens, the fluid becomes proteinaceous, and the surrounding tissues become edematous (248) (Fig. 4-119). The fluid usually is hypointense to hyperintense on T1-weighted images and hyperintense on T2-weighted images (79) (Fig. 4-120). Second branchial cleft cysts characteristically occur anterior to the sternocleidomastoid muscle near the angle of the mandible, although these cysts may arise anywhere along the path of a second branchial cleft fistula or sinus and even expand posteriorly into the posterior triangle of the neck (135,155,367). Extension of the mass between the external and internal carotid arteries is very suggestive of a second branchial cleft cyst regardless of its relation to the sternocleidomastoid muscle (321). Bronchial cysts are derived from buds of the foregut that normally differentiate into the tracheobronchial tree (202,246,355,367). Trapped respiratory epithelium or possibly migrating primordial cells secrete mucus within the cyst and eventually enlarge the lesion. These cysts rarely can present in the low neck. On MR images, a fluid signal intensity is observed (Fig. 4-121). Occasionally, these cysts may contain air (367). Tracheoesophageal cysts also are derived from foregut (367). Ranulas are sublingual salivary gland retention cysts and, by definition, arise in the sublingual space. Simple ranulas are confined to the floor of the mouth. Plunging ranulas dehisce either through or posterior to the mylohyoid muscle and spill into submandibular and parapharyngeal

A

B Figure 4-112 Thyroglossal duct cyst. Enhanced computed tomography scan (A) shows a hypodense left neck lesion (asterisk) located within the paralaryngeal strap muscles (m). This appearance resembles a “snake swallowing an egg” (B). Thyroglossal duct cyst (asterisk), strap muscles (m). (From Brown J, Wippold FJ 2nd. Practical MRI: a teaching file. New York: Lippincott-Raven Press, 1996:200–220, with permission.)

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 203

Neck

A

203

B Figure 4-113 Thyroglossal duct cyst. Enhanced computed tomography at the level of the hyoid bone (A) shows a lateral cystic lesion (asterisk) notching the inner surface of the hyoid (arrow). Inferiorly, at the level of the pyriform sinuses (B), the lesion (asterisk) is embedded within the paralaryngeal strap muscles (m).

spaces (62,67,355,398) (Fig. 4-122). MRI is ideally suited for characterizing floor of the mouth cystic lesions (424). The MRI appearance of ranulas is usually cystic; however, signal intensity varies with the protein or hemorrhagic contents of the ranula (245,398). Epidermoids may resemble ranulas (62,67). Epidermoids and dermoids are epithelial-lined inclusion cysts and may present in the floor of the mouth, orbit, or nasal region. Both epidermoids and dermoids contain stratified squamous epithelium; however, dermoids additionally contain skin appendages such as hair and sebaceous glands (202,352). Both epidermoids and dermoids are entirely derived from the ectoderm (352). These cysts also are filled with debris and cholesterol from the breakdown of keratin. Dermoids are virtually always midline and frequently occur in the sublingual space or submental region (see Fig. 4-123) (245). Epidermoids may be either

midline or lateral. Occasionally, pitting of the overlying skin is observed on palpation. Teratomas also may be cystic and demonstrate mixed tissue elements on imaging (367). On CT images, epidermoids are low density and homogeneous. On MR images, the lesions are typically hypointense on T1-weighted images and hyperintense on T2weighted images (117,395). Lipid within dermoids is caused by the breakdown of sebaceous debris and desquamation and contributes to frequently demonstrated hyperintensity on T1-weighted images (352). Observed calcification may be dystrophic or secondary to the production of enamel, also derived from ectoderm (352). Lymphatic malformations are considered low-flow vascular malformations, in contradistinction to high-flow lesions

Figure 4-115 Figure 4-114

Thyroglossal duct cyst. T1-weighted magnetic resonance image demonstrating a mildly hyperintense midline lesion (arrow) notching the dorsal surface of the hyoid bone (arrowheads).

Infected laryngeal mucocele. Enhanced computed tomography reveals a cystic mass with peripheral enhancement (asterisk) straddling the thyrohyoid membrane and compromising the airway. The skin is thickened (arrows), the subcutaneous fat is infiltrated, and the platysma muscle is thickened (arrowheads) as a result of inflammation.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 204

204

Chapter 4

Figure 4-117 Second branchial cleft cyst. Computed tomography reveals a large thin-walled cystic mass (asterisk) anterior to the left sternocleidomastoid muscle (m) and posterior to the left submandibular gland (G). Figure 4-116 Infected laryngeal mucocele. A well-circumscribed cystic mass (asterisk) perforates the thyrohyoid membrane and demonstrates both intralaryngeal and extralaryngeal components. The surrounding fat planes are obscured. The markedly compromised airway has required intubation (arrowhead). Calculi (arrows) are incidentally noted within the right submandibular gland.

such as arteriovenous malformations (104). Lymphatic malformations arise when formative lymph sacs become isolated from the central draining venous system during embryologic development and then continue to enlarge

(104). These lesions are most commonly cystic and carry such names as cystic hygromas, cavernous lymphangiomas, capillary lymphangiomas, and vasculolymphatic malformations (77,202,304,461). Microcystic lesions may infiltrate muscle and skin, whereas macrocystic malformations have no cutaneous manifestations (104). Lesions may dramatically expand in size due to hemorrhage or infection (104). Approximately 75% of these lesions are found in the neck, and frequently within the posterior triangle and posterior cervical

A

B Figure 4-118 Branchial cleft cyst. A well-defined, homogeneous, fluid-filled mass (arrowheads) anterior to the left sternocleidomastoid muscle is identified on magnetic resonance images. It is hypointense on T1-weighted (A) and hyperintense on T2-weighted (B) images. The surrounding fat and soft tissues are normal. (From Brown J, Wippold FJ 2nd. Practical MRI: a teaching file. New York: Lippincott-Raven Press, 1996:200–220, with permission.)

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 205

Neck

Figure 4-119

Infected second branchial cleft cyst. Computed tomography scan in a patient with a tender right neck mass demonstrates an enhancing thick-walled cystic mass (asterisk) anterior to the right sternocleidomastoid muscle (m).

space (see Fig. 4-89) (337,446). Almost 70% of these lesions present at birth, and 90% have presented by 2 years of age (Fig. 4-124). On imaging, these lesions may be solitary or cystic without a definite cyst wall. Fluid on T1-weighted images may be hypointense or hyperintense depending on lipid, blood, or protein content (Fig. 4-125). Cystic hygromas frequently infiltrate along tissue planes (349,355,367). Nerve sheath tumors of the neck may involve the vagus, hypoglossal, or cervical roots (367). Schwannomas and

205

neurofibromas are the two most commonly encountered nerve sheath tumors in clinical practice (146). Both schwannomas and neurofibromas are composed principally of Schwann cells and are difficult to differentiate radiographically. Histologically, schwannomas usually have two cell populations. The Antoni A cells consist of compact spindle cells arranged in bundles separated by collagen. The Antoni B cells have a loose myxoid matrix (292). An MRI target sign has been described in 50% to 70% of nerve sheath tumors. This appearance consists of a central hypointensity surrounded by a hyperintense rim on T2-weighted and enhanced T1-weighted images. The pattern corresponds histologically to central fibrocollagenous material encircled by peripheral myxomatous tissue (see Fig. 4-24). The target sign may be absent when the tumor undergoes necrosis, hemorrhage, or cystic change (382,417). Central degeneration may produce an imaging appearance of cystic change in over 30% of lesions (Fig. 4-126) (355). Lipomas are not cystic lesions, but may mimic a cyst on CT if admixed with soft tissue strands. The fat signature on imaging is diagnostic. Over 10% of lipomas occur in the head and neck. Almost 10% are multiple. The posterior triangle is the most common location, although lipomas can occur in the parotid space, parapharyngeal space, buccal space, anterior neck, and oral cavity (Fig. 4-89 and 4-127) (94,369). Patients are usually in the sixth or seventh decade at presentation. Lipomas tend to enlarge during periods of weight gain, but usually do not recede during dieting. The overlying skin may be cool to clinical palpation.

A

B Figure 4-120

Infected branchial cleft cyst. A: Computed tomography image at the level of the hyoid bone (H) demonstrates a thick-walled, partially cystic mass (large arrows) anterior to the sternocleidomastoid muscle (SCM). The surrounding fascial planes are thickened; left internal carotid artery (arrowhead); left internal jugular vein (small arrow); submandibular gland (G). B: On T2-weighted magnetic resonance image at approximately the same level, the central fluid component of the mass is much more hyperintense than the thickened wall; left carotid artery (arrowhead); left jugular vein (small arrow); sternocleidomastoid muscle (SCM); submandibular gland (G).

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 206

206

Chapter 4

Figure 4-121

Bronchial cyst. T2-weighted magnetic resonance image demonstrates a large, midline hyperintense cystic neck mass (asterisk) in a young child. The mass displaces the airway anteriorly and straddles both the suprahyoid and infrahyoid regions of the neck. Differential diagnosis includes: cystic hygroma and dermoid. (From Brown J, Wippold FJ 2nd. Practical MRI: a teaching file. New York: Lippincott-Raven Press, 1996:200–220, with permission.)

Thrombosed internal jugular veins also may appear cystic on imaging. The surrounding tissues may be edematous in cases of phlebitis (5,304). Recognition of the tubular configuration of the lesion usually secures the diagnosis (Fig. 4-128). Thyroid cysts arise from the thyroid gland. The colloid cyst is firm and may expand relatively rapidly. The hemorrhagic

Figure 4-122

Plunging ranula. Enhanced computed tomography image demonstrating a bilobed, thin-walled, cystic mass originating in the left sublingual space (arrowheads) and extending into the left submandibular space (arrows). Mylohyoid muscle (m), geniglossus muscles (g).

cyst results from bleeding into an adenoma and usually is painful (355,367). The imaging appearance is nonspecific and may resemble a parathyroid cyst, cervical thymic cyst, or a cyst developing in thyroid carcinoma (Fig. 4-129). Diagnosis may require sonographically guided aspiration. Parathyroid cysts arise in the parathyroid glands. Over 60% involve the inferior parathyroid glands. Although usually asymptomatic, these benign lesions may cause hoarseness due to compression of the recurrent laryngeal nerve. Functioning tissue can cause hyperparathyroidism. Nonfunctioning cysts most commonly affect women and occur in the inferior glands, whereas functioning cysts are slightly more common in men and may arise anywhere in the neck. The imaging appearance of parathyroid cysts is nonspecific. Cysts are usually thin walled and lateral to the thyroid gland (Fig. 4-130). Hemorrhage may be evident. Differentiation from thyroid cysts may be difficult. Aspiration of fluid from within the cyst usually reveals elevated parathormone and is useful in establishing the site of origin (355,367). Cervical thymic cysts develop from focal areas of degeneration of the thymus or from remnants of the thymopharyngeal ducts (202). These latter structures arise from the third and fourth pharyngeal pouches, migrate inferiorly in the neck, and eventually become the thymus. Nearly 90% of patients present with painless neck swelling. These lesions are closely related to the carotid sheath and may arise from the angle of the mandible to the mediastinum (446). The nonspecific imaging appearance is influenced by cholesterol and debris that may be present within the fluid and cyst walls (355,367). Tornwaldt cysts arise from notochordal remnants and are seen in 4% of the normal population. These proteinaceous fluid-filled cysts are typically found in the midline nasopharynx (Fig. 4-131). Tornwaldt cysts are asymptomatic unless complicated by infection (248).

Figure 4-123

Dermoid cyst. Enhanced computed tomography scan demonstrates a large, thin-walled midline cyst (asterisk) within the floor of the mouth. The water-density contents are homogeneous. Location suggests the correct diagnosis. Differential diagnosis includes epidermoid, ranula, and thyroglossal duct cyst.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 207

Neck

A

207

B

Figure 4-124

Cystic hygroma. A huge cystic mass (asterisk) covers the entire right side of this infant’s face. The lesion is hyperintense on coronal proton density–weighted (A) and T2-weighted Magnetic resonance (B) images. Septations (arrowheads) within the hygroma divide the lesion into loculated compartments.

POSTTREATMENT NECK Imaging evaluation of the treated neck is challenging because the usual imaging landmarks are distorted or absent. Nevertheless, imaging is useful in confirming recurrent disease, demonstrating extent of the lesion, and directing biopsies (83,84,118,156). Treatment may involve surgery, radiation, chemotherapy, immunotherapy, or combined modalities. Surgical approaches can be divided into surgery

for the primary lesion, such as laryngectomy for laryngeal cancer, adjunctive surgery for spread of disease, such as neck dissection for cervical lymphadenopathy, and reconstructive surgery, such as construction of myocutaneous flaps. Surgery for the primary lesion also can be subdivided into conservative procedures and radical procedures. Conservative procedures for primary laryngeal lesions attempt to preserve voice and some laryngeal function and include simple cordectomy, horizontal partial laryngectomy

A

B Figure 4-125 Cystic hygroma. A: On T1-weighted magnetic resonance (MR) image, a multiloculated mass (M), slightly lower in signal intensity than fat, is seen in the posterior cervical space, displacing the sternocleidomastoid muscle laterally (small arrowheads). The mass infiltrates between adjacent soft tissues and separates the internal carotid artery (large arrowhead) and internal jugular vein (arrow). B: The mass (M) is very hyperintense on the T2-weighted MR image, reflecting its fluid nature.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 208

208

Chapter 4

Figure 4-126

Vagus nerve schwannoma. Computed tomography scan demonstrates a large left suprahyoid soft tissue neck mass (arrows) arising in the poststyloid compartment of the parapharyngeal space. The lesion displaces the carotid artery and jugular vein anteriorly and encroaches on the airway. Cystic components (arrowheads) are evident within the mass.

(supraglottic partial laryngectomy, supracricoid partial laryngectomy), and vertical partial laryngectomy (frontolateral partial laryngectomy, frontoanterior partial laryngectomy, hemiglottolaryngectomy). Successful conservative laryngeal surgery is contingent on the preservation of the cricoid cartilage, one superior and one inferior laryngeal nerve, and one cricoarytenoid articulation (126). Assessment of possible subglottic extension, anterior commissure

Figure 4-127 Lipoma. Computed tomography (CT) scan shows a large homogeneous fat-density mass (asterisk) in the left posterior triangle of the neck. The lesion displaces subjacent muscles and deforms the contour of the neck. CT attenuation values distinguish lipomas from water-density cysts.

Figure 4-128 Thrombosed internal jugular vein. On the enhanced computed tomography image, the left internal jugular vein (arrow) fails to fill with contrast. The lumen is hypodense. Surrounding soft tissue planes are preserved, indicating a lack of radiologically evident inflammation. The tubular configuration of the lesion, established with several contiguous slices, confirms the diagnosis. Internal carotid arteries (C), right internal jugular vein (J).

involvement, superior vertical extension, and arytenoid cartilage involvement is essential for surgical planning (69). Supraglottic laryngectomy is performed for selected lesions of the epiglottis, unilateral arytenoid, aryepiglottic fold, and false vocal cord (249,273,441). In this procedure, the epiglottis, aryepiglottic folds, false vocal cords, superior thyroid ala, pre-epiglottic space, and often a substantial portion of the hyoid bone are removed. For tumors involving the superior portion of the arytenoid cartilage or medial pyriform sinus, the arytenoid and affected pyriform sinus mucosa also are removed (273). Contraindications for supraglottic laryngectomy include invasion of the cricoid or thyroid cartilage, involvement of the apex of the pyriform sinus or postcricoid region, substantial involvement of the base of the tongue, extension into the glottis, and impairment of cord mobility and arytenoid fixation or bilateral arytenoid involvement (50). On postsurgical imaging, the anatomy below the level of the laryngeal ventricle is undisturbed. Above this level, the arytenoids are prominent (441) (Fig. 4-132). Loss of the normal fat signal in the upper outer margin of the thyroarytenoid muscle signifies tumor invasion of the cord and precludes this surgical procedure (50). Complications of supraglottic laryngectomy include infection, wound slough, fistula, hemorrhage, aspiration, and carotid artery exposure (249,273). Because of the high prevalence of lymph node metastases, a neck dissection often is included in the procedure. Vertical hemilaryngectomy may be recommended for true vocal cord lesions limited to one cord with normal mobility. Vertical hemilaryngectomy involves resection of the ipsilateral true vocal cord, laryngeal ventricle, false vocal cord, and thyroid cartilage ala. Up to one third of the contralateral cord and a portion of the cricoid cartilage may be resected in cases of limited cancer spread; however, the procedure is usually contraindicated if more than 33% of

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 209

Neck

209

A

B Figure 4-129

Nonspecificity of hypodense thyroid lesions on computed tomography images. A: A relatively low-attenuation mass (arrows) resulting from nodular hyperplasia is seen in the right lobe of the thyroid gland; common carotid artery (C); internal jugular vein (J); left lobe of thyroid (Th). B: Another patient presents with a similar appearing low-attenuation nodule (arrow), due to metastatic adenocarcinoma, within the right lobe of the thyroid gland; carotid artery (C); jugular vein (J).

the contralateral cord is involved (37,84,431). Extensive subglottic or supraglottic spread, involvement of the cricoarytenoid joint, interarytenoid area, or contralateral arytenoid, and thyroid cartilage invasion preclude use of this procedure. A pseudocord may be reconstructed using omohyoid or sternohyoid muscle and thyroid perichondrium (84). Postsurgical imaging is variable but demonstrates absence of the ipsilateral aryepiglottic fold, asymmetry of the thyroid cartilage, and presence of soft tissue within the preepiglottic space (Fig. 4-133). The reconstructed pseudocord is usually straight or concave, and the residual contralateral cord is convex. Well-defined masses within the preepiglottic space, bulging of the pseudocord, or masses within the contralateral cord are considered signs of recurrence (84).

Figure 4-130 Parathyroid cyst. A large, low-attenuation mass (M) displaces the trachea (T) to the right, and the left common carotid artery (arrow) and internal jugular vein (arrowhead) to the left; scalenus anterior muscle (S). Preoperative diagnosis was a thyroid cyst. No clinical manifestations of parathyroid hyperfunction were present.

Total laryngectomy is indicated for cancers that extensively involve the preepiglottic and paraglottic spaces, that invade the cricoid or thyroid cartilages, and that involve the subglottic region. In this procedure, the entire larynx is removed and the patient is left with a tracheostomy and a constructed neopharynx (83,326,441). Remnants of thyroid tissue not removed at the time of the original laryngectomy may appear as unilateral or bilateral asymmetric hyperdense masses on CT (433). Imaging demonstrates the neopharynx as a tube extending from the base of the tongue to the cricopharyngeus muscle and the absence of the laryngeal cartilages (Figs. 4-134 and 4-135) (441). Imaging signs denoting tumor recurrence include effacement of fat planes surrounding the neopharynx, neurovascular bundles, and sternocleidomastoid muscles, infiltration of muscle, and thickening of the tracheostomy wall and peristomal soft tissues (83). Pharyngocutaneous fistulae may complicate total laryngectomy, especially in patients who have received preoperative radiation or concomitant radical neck dissection or who have coexisting systemic disease (285). One of the most commonly encountered adjunctive surgical procedures is the radical neck dissection. The radical neck dissection is indicated for clinically apparent lymphadenopathy found during the evaluation and staging of a primary lesion, for metastatic lymphadenopathy in patients without a known primary lesion, and for clinically N0 necks in patients with primary tumors that have a marked tendency for lymph node spread. This surgery entails the removal of all ipsilateral lymph nodes from levels I through V, the submandibular gland, the internal jugular vein, the sternocleidomastoid muscle, and the spinal accessory nerve and may further require removal of the sternohyoid muscle, sternothyroid muscle, anterior and posterior bellies of the digastric muscle, and ipsilateral thyroid gland (Fig. 4-136) (359,370,437,441). The sacrifice of the spinal accessory nerve causes atrophy of the trapezius muscle (the

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 210

210

Chapter 4

B

A Figure 4-131

Tornwaldt cyst. Proton density–weighted (A) and T2-weighted (B) magnetic resonance images demonstrate a hyperintense midline nasopharyngeal cystic mass (arrows).

A

C

B

Figure 4-132 Normal anatomy after supraglottic subtotal laryngectomy and left radical neck dissection. A: base-oftongue level; residual hyoid bone (arrows); base of tongue (B); left internal carotid artery (arrowhead). B: midneck level. A portion of the thyroid cartilage has been removed. The pseudo-false cords (arrows) are seen extending anterolaterally from the arytenoid cartilages (a); laryngeal vestibule (V); pyriform sinus (small arrowhead). C: low-neck level. The axis of the airway (dotted line) is tilted slightly away from the side of the radical neck dissection; common carotid artery (C); right internal jugular vein (J); thyroid gland (Th); esophagus (E).

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 211

211

Neck

A

B Figure 4-133

Normal anatomy after left vertical hemilaryngectomy. A: level of superior cornua of thyroid cartilage (t); preepiglottic space (PES), with diminution in fat content; hyoid bone (H); residual aryepiglottic fold on nonoperated side (white arrow). B: Level of glottis. Note tilt of airway toward the operated side; residual left thyroid lamina (arrowhead); surgically created pseudocord (p); undersurface of residual right true vocal cord (v).

A

C

B

Figure 4-134 Normal anatomy after total laryngectomy and modified left neck dissection. A: Base-of-tongue level. B: Mid-neck level. C: Tracheostomy level. Appearance of multiple lumina in the esophagus (small arrowheads) is secondary to apposition of the anterior and posterior walls by mucus threads; base of tongue (B); submandibular gland (G); right internal jugular vein (J); neopharynx (N); sternocleidomastoid muscle (SCM); internal carotid arteries (small black arrows); common carotid arteries (large black arrows); external jugular vein (white arrow); scalenus anterior muscle (large arrowhead); tracheostomy (white arrowhead).

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 212

212

Chapter 4

Figure 4-136 Figure 4-135

Total laryngectomy. T1-weighted sagittal magnetic resonance image demonstrating the typical appearance of a total laryngectomy. The larynx, with its characteristic imaging landmarks, has been removed. The neopharynx is collapsed (arrowheads) and a tracheostomy is evident (arrow).

sternocleidomastoid is dissected as part of the surgery), and the ipsilateral levator scapulae muscle assumes the functional role for these muscles. The resultant levator scapulae hypertrophy should not be mistaken for tumor recurrence (335). With increased clinical understanding of the behavior of certain cancers, a variety of modified radical neck dissections have been implemented to tailor surgery by preserving specific nonnodal structures, such as the internal jugular vein, sternocleidomastoid muscle, and spinal accessory nerve (359,370). This approach lessens postoperative morbidity. Selective neck dissections leave certain low-risk nodal groups intact. Alternatively, more extensive disease demands extended radical neck dissections that remove nodal groups in addition to groups I to V as well as any nonnodal structures deemed at risk (437). Identification of the specific procedure on imaging is greatly facilitated by consulting the surgeon. Systematic identification of all neck structures as a means of recognizing missing elements is more helpful than attempting to merely label the surgery. Neck reconstruction following major dissections attempts to restore cosmetic and functional utility. Skin grafts and more substantial flaps are used to fill surgically created anatomic defects and cover vital structures, such as exposed carotid arteries, and to create structures such as neopharyngeal tubes. Flaps may consist of skin and subcutaneous tissues (cutaneous flaps) as well as muscle (myocutaneous flaps) and occasionally bone (osteocutaneous or osteomyocutaneous flaps) (76,279,370). Flaps may be derived from local tissues and rotated into place preserving a vascular pedicle or donated from a distant site (free

Radical neck dissection. Enhanced computed tomography scan reveals surgical absence of the left jugular vein and sternocleidomastoid muscle. The left carotid artery (c) is preserved. Right sternocleidomastoid muscle (M), right internal jugular vein (j), common carotid arteries (c).

flaps) with surgical anastomoses of vascular pedicles (76,370,441). On imaging, the reconstructed neck is usually grossly distorted with a prominent fatty mass. The vascular pedicle or incorporated breast tissue may appear as a soft tissue density and should not be mistaken for recurrent cancer. Hematomas, lymphoceles, and abscesses also complicate imaging interpretation by presenting as masses (279). Tumor recurrence within the flap bed most commonly manifests on CT as a focal, dense mass, often with necrosis. Diffusely distorted soft tissue planes also herald recurrence; however, early detection may be confounded by superimposed tissue changes caused by radiation (173). Postsurgical growth of residual lymph nodes is an ominous sign. Consultation with the surgeon is extremely important in verifying the treatment and in communicating any clinically suspicious imaging findings (173). Potential mimics of myocutaneous flaps on CT include massive obesity (especially in the patient scanned with the head poorly positioned, causing asymmetry of anatomic structures) and lipomas. Regardless of the surgical procedure, hemorrhage and edema are usually evident on imaging during the early postoperative period and persist for 4 to 6 weeks. A baseline study is therefore best postponed for at least a month to avoid potentially confusing soft tissue masses. Scanning at intervals of 4 months to 6 months for the first 1 to 3 years is then recommended (81,370). Yearly examinations are then performed unless a change in clinical status prompts a more timely scan (370). Unfortunately, cancers of the neck have an overall high rate of recurrence, and detection on imaging is crucial for direction of secondary

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 213

Neck

213

and salvage therapies. In 25% of patients, imaging is the only method to adequately detect recurrent tumor. In 30% of patients, imaging demonstrates more extensive disease than is clinically suspected (81,154). Typically, cancer recurrence within the first 2 months after surgery is unusual if the specimen margins were free of tumor (370). With time, the bulky fatty appearance of the flap diminishes as atrophy and scar develop. Tissue planes become more defined but remain somewhat obscured compared with preoperative images. The development of a dense soft tissue mass usually indicates recurrent tumor (see Fig. 4-64). Scar tends to be less dense on CT but may be difficult to differentiate from recurrent neoplasm without biopsy. MRI usually is unable to reliably distinguish recurrent tumor from edema, radiation necrosis, or inflammatory lesions (291). Fibrosis tends to be hypointense on T2-weighted images, whereas recurrent tumor is hyperintense (81). Growth of any soft tissue density on imaging is another sign of recurrence, whereas scar remains stable or diminishes with time (370). Radiation therapy is frequently indicated in the management of neck cancers (394). Soft tissue changes may be observed on imaging with doses of 6,500 cGy to 7,000 cGy (50,262). On imaging, the skin and platysma are thickened. Increased capillary permeability contributes to interstitial edema, which causes stranding and hyperdensity of the subcutaneous fat on CT (Fig. 4-137). Fluid may accumulate within the mastoid cells and is best appreciated on T2-weighted MR images (275). Radiation ports that extend to the floor of the mouth may also cause eventual fibrosis of the muscles of mastication with sufficiently high doses and may result in trismus (28). The muscles become diffusely hyperintense on T2-weighted MR images and may enhance following contrast administration. With doses approaching 7,000 cGy, skin ulceration becomes increasingly likely (397).

The major and minor salivary glands are especially sensitive to radiation, and 1,000 cGy to 2,000 cGy cause mucositis (397). On CT images, the major salivary glands enhance and then eventually atrophy (262,275) (see Fig. 4-137). Three-dimensional radiation port planning, which permits shielding of the parotid glands, ameliorates posttherapy xerostomia (96). Radiation directed to the larynx and pharynx causes generalized thickening of normal structures and increased attenuation of the fat-containing pre-epiglottic and paralaryngeal spaces (57,275,397). Thickening of the epiglottis, aryepiglottic folds, false vocal cords, and the true cords is usually diffuse, symmetric, and dose dependent (57,262, 275) (Fig. 4-138). The mucosa and submucosa of the subglottis also are thickened, and therefore early detection of infiltrating cancer is hindered (262). The mucosal surfaces may enhance because of mucositis. Contraction of the thyrohyoid membrane may cause the hyoid bone and thyroid cartilage to appear on the same imaging slice (see Fig. 4-65). Radiation-induced soft tissue fibrosis may also contribute to dysphagia (28). Higher doses of radiation weaken membranes and cartilage and may contribute to chondronecrosis and cartilage collapse (2,8,50). Radiated cartilage also is more prone to infection and may develop perichondritis if biopsied (28,71). High doses of radiation may also cause soft tissue necrosis and pharyngocutaneous fistula. The latter may be exacerbated with continued tobacco and alcohol consumption (28). Other organs may be affected by neck radiation. The adjacent skeleton may demonstrate hyperintense marrow signal on T1-weighted MR images because of post-therapy marrow fat infiltration, the mandible may succumb to osteoradionecrosis, and rarely, the radiated bone may give rise to sarcomas (275,397). Osteoradionecrosis involves the devitalization of radiated bone and results from

Figure 4-137 Radiation-induced soft tissue changes and sialadenitis. Enhanced computed tomography in a patient receiving radiation to the right neck. The subcutaneous fat of the right neck is edematous and stranded (arrows), the platysma is thickened (arrowheads), and the right submandibular salivary gland (asterisk) is enlarged and enhanced. Compare to the nonradiated left neck.

Figure 4-138 Radiation-induced laryngeal edema. Enhanced computed tomography image demonstrates symmetric supraglottic swelling, irregular enhancement, and airway narrowing following radiation therapy for glottic carcinoma.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 214

214

Chapter 4

Figure 4-140 Teflon injection, true vocal cord. Enhanced T1weighted magnetic resonance image shows a hypointense structure within the right true vocal cord and paraglottic space corresponding to Teflon (arrowhead). The injected cord bulges toward the midline.

Figure 4-139 Cancer recurrence. Combined positron emission tomography/computed tomography (PET/CT) scan demonstrating PET images of hypermetabolic cancer recurrence in the right tongue base and right neck (arrows) superimposed on the axial CT scan of the corresponding level.

vascular damage and direct injury to osteoblasts augmented by xerostomia and resultant dental caries. The involved bone exhibits focal demineralization, disorganized trabeculae, cortical thickening, patchy sclerosis, and, occasionally, pathological fractures (397). Periosteal reaction is unusual and should prompt consideration of infection. Although associated soft tissue mass may suggest tumor recurrence, care must be exercised in the diagnosis of radiated tissue adjacent to necrotic bone because soft tissue thickening and enhancement unrelated to tumor have been described (63). In addition to the mandible, osteoradionecrosis may also involve the skull base and hyoid bone (28). Postradiation sarcomas are rare and usually appear after a latent period measured in decades. Because of the age of the typical neck cancer patient, the associated medical debilities from chronic smoking, and the unfortunately high rate of fatal primary tumor recurrence, sarcomas are not usually seen. Cranial neuropathies may also develop following head and neck radiation. Although the hypoglossal nerve is most commonly affected, the vagus and spinal accessory nerves may also be involved. Neuropathies are usually delayed, appearing months to years after treatment, and are thought to be the result of reactive fibrosis or vascular in-

sufficiency (28,33,185,327). Accelerated atherosclerotic disease of the carotid arteries may occur in radiated necks and may manifest as long segment stenoses corresponding to the treatment field on MR or CT angiography (28). Aneurysms or frank occlusions may also be seen. Imaging may reveal local radiation failure several months before the failure is clinically recognized (163). Differentiating postradiation changes from recurrent tumor may be difficult on CT and MRI, however. Some investigators have suggested that fibrosis tends to be hypointense on T1-weighted and T2-weighted images, whereas tumor recurrence has higher signal (139,397). Other authors have been less encouraged because of the potentially confounding signal changes caused by hemorrhage and edema (45). The complete or near complete resolution of the primary lesion on imaging is one of the best criteria for a successful radiation response. Residual mass of greater than 50% of the original tumor volume usually connotes a therapy failure (261). Post-treatment development of tongue atrophy or sternocleidomastoid and trapezius atrophy on CT or MR imaging may signal recurrence; however, a delayed effect of radiation should be considered if no mass is found (33,185,327). The appearance of new focal masses and ulcerations on barium pharyngography support the diagnosis of recurrence (296). Other modalities, such as FDG-positron emission tomography (PET) or FDG-SPECT, may ultimately prove more useful than conventional cross-sectional imaging (57). Combined PET/CT is a new technique in which functional PET data are superimposed on near-simultaneously acquired anatomic CT data. By combining the superior sensitivity of PET with the spatial resolution of CT, tumor localization can be confidently identified (Fig. 4-139).

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 215

Neck

Figure 4-141

Teflon injection, true vocal cord. computed tomography demonstrates greater density (arrowhead) of teflon injected into the left true vocal cord and paraglottic space. The mucosal surface of the injected cord is pushed toward the midline to oppose the contralateral cord during phonation.

A variety of synthetic materials may be inserted into the larynx and neck. Teflon injected into the paralaryngeal space adds bulk to a paralyzed or reconstructed vocal cord (Figs. 4-140 and 4-141). Stents may be placed to ensure a patent airway in patients with laryngeal stenosis. Speaking tubes and valves are designed to help the laryngectomy patient use the esophagus in speech production. Tracheostomy tubes are inserted into the midcervical airway in the patient with laryngeal stenosis, acutely compromised airway, or laryngectomy (71). Some of these materials may be noted incidentally on scans in patients who have failed to mention past surgical procedures and should not be misinterpreted as masses or traumatically induced foreign bodies.

REFERENCES 1. Aasen S, Kolbenstvedt A. CT appearances of normal and obstructed submandibular duct. Acta Radiol 1992;33: 414–419. 2. Adams GL. Malignant neoplasms of the hypopharynx. In: Cummings CW, Frederickson JM, Harker LA, et al., eds. Otolaryngology– head and neck surgery. St. Louis: CV Mosby, 1993;1955–1973. 3. Agha FP. Recurrent laryngeal nerve paralysis: a laryngographic and computed tomographic study. Radiology l983;148:149–155. 4. Ahmadi J, Hinton DR, Segall HD, et al. Dural invasion by craniofacial and calvarial neoplasms: MR imaging and histopathologic evaluation. Radiology 1993;188:747–749. 5. Albertyn LE, Alcock MK. Diagnosis of internal jugular vein thrombosis. Radiology 1987;162:505–508. 6. Ali YA, Saleh EM, Mancuso AA. Does conventional tomography still have a place in glottic cancer evaluation? Clin Radiol 1992;45:114–119. 7. Anderson CM, Gooding GAW, Lee RE. Thyroid cyst mistaken for carotid pseudoaneurysm by MR angiography: case report. Clin Imaging 1992:16:198–200.

215

8. Anzai Y, Blackwell KE, Hirschowitz SL, et al. Initial clinical experience with dextran-coated superparamagnetic iron oxide for detection of lymph node metastases in patients with head and neck cancer. Radiology 1 994;192: 709–715. 9. Anzai Y, Lufkin RB, Jabour BA, et al. Fat-suppression failure artifacts simulating pathology on frequency-selective fat-suppression MR images of the head and neck. AJNR Am J Neuroradiol 1992;13:879–884. 10. Anzai Y, McLachlan S, Morris M, et al. Dextran-coated superparamagnetic iron oxide, an MR contrast agent for assessing lymph nodes in the head and neck. AJNR Am J Neuroradiol 1994;15: 87–94. 11. Archer CR, Sagel SS, Yeager VL, et al. Staging of carcinoma of the larynx: comparative accuracy of CT and laryngography. AJR Am J Roentgenol 1981;136:571–575. 12. Archer CR, Yeager VL. Evaluation of laryngeal cartilages by computed tomography. J Comput Assist Tomogr 1979;3:604–611. 13. Archer CR, Yeager VL, Herbold DR. Improved diagnostic accuracy in laryngeal cancer using a new classification based on computed tomography. Cancer 1984;53:44–57. 14. Archer CR, Yeager VL, Herbold DR. Improved diagnostic accuracy in the TNM staging of laryngeal cancer using a new definition of regions based on computed tomography. J Comput Assist Tomogr 1983;7:610–617. 15. Aspestrand F, Kolbenstvedt A, Boysen M. Carcinoma of the hypopharynx: CT staging. J Comput Assist Tomogr 1990;14:72–76. 16. Auffermann W, Clark OH, Thurnher S, et al. Recurrent thyroid carcinoma: characteristics on MR images. Radiology 1988;168: 753–757. 17. Avrahami E. Panic attacks during MR imaging: treatment with i.v. diazepam. AJNR Am J Neuroradiol 1990;11:833–835. 18. Babbel RW, Harnsberger HR. The parapharyngeal space: the key to unlocking the suprahyoid neck. Semin Ultrasound CT MR 1990;11:444–459. 19. Baker SR, Latack JT. Magnetic resonance imaging of the head and neck. Otolaryngol Head Neck Surg 1986;95:82–89. 20. Balaban RS, Ceckler TL. Magnetization transfer contrast in magnetic resonance imaging. Magn Reson Q 1992;8:116–137. 21. Barakos JA, Dillon WP, Chew WM. Orbit, skull base, and pharynx: contrast-enhanced fat suppression MR imaging. Radiology 1991;179:191–198. 22. Bass IS, Haller JO, Berdon WE, et al. Nasopharyngeal carcinoma: clinical and radiographic findings in children. Radiology 1985;156:651–654. 23. Batsakis JG, Raymond AK. Cartilage tumors of the larynx. South Med J 1988;81: 481–484. 24. Batsakis JG. Neoplasms of the larynx. In: Batsakis JG. Tumors of the head and neck: clinical and pathologic considerations, 2nd ed. Baltimore: Williams & Wilkins, l979:200–225. 25. Batsakis JG. Parenchymal cysts of the neck. In: Batsakis JG. Tumors of the head and neck: clinical and pathological considerations, 2nd ed. Baltimore: Williams & Wilkins, 1979:233–251. 26. Becker M. Larynx and hypopharynx. Radiol Clin North Am 1998;36:891–920. 27. Becker M, Moulin G, Kurt A-M, et al. Non-squamous cell neoplasms of the larynx: radiologic-pathologic correlation. Radiographics 1998;18:1189–1209. 28. Becker M, Schroth G, Zbären P, et al. Long-term changes induced by high-dose irradiation of the head and neck region: imaging findings. Radiographics 1997;17:5–26. 29. Becker M, Zbären P, Delavelle J, Kurt A-M, et al. Neoplastic invasion of the laryngeal cartilage: reassessment of criteria for diagnosis at CT. Radiology 1997;203:521–532. 30 Becker M, Zbären P, Laeng H, et al. Neoplastic invasion of the laryngeal cartilage: comparison of MR imaging and CT with histopathologic correlation. Radiology 1995;194:661–669. 31. Becker RL, Norfray JF, Teitelbaum GP, et al. MR imaging in patients with intracranial aneurysm clips. AJNR Am J Neuroradiol 1988;9:885–889. 32. Bent JP 3rd, Porubsky ES. The management of blunt fractures of the thyroid cartilage. Otolaryngol Head Neck Surg 1994;110: 195–202. 33. Berger PS, Bataini JP. Radiation-induced cranial nerve palsy. Cancer 1977;40:152–155.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 216

216

Chapter 4

34. Berkovitz BKB, Moxham BJ. The larynx. In: Berkovitz BKB, Moxham BJ. A textbook of head and neck anatomy. Barcelona: Wolfe Publishing, 1988;332–360. 35. Bhalla M, Grillo HC, McLoud TC, et al. Idiopathic laryngotracheal stenosis: radiologic findings. AJR Am J Roentgenol 1993; 161:515–517. 36. Bielamowicz SA, Storper IS, Jabour BA, et al. Spaces and triangles of the head and neck. Head Neck 1994;16:383–388. 37. Biller HF, Ogura JH, Pratt LL. Hemilaryngectomy for T2 glottic cancers. Arch Otolaryngol 1971;93:238–243. 38. Bischof TP, von Schulthess GK. Imaging techniques in thyroid and parathyroid disease. Curr Opin Radiol 1990; 2:122–127. 39. Boey HP, Cunningham MJ, Weber AL. Central nervous system imaging in the evaluation of children with true vocal cord paralysis. Ann Otol Rhinol Laryngol 1995;104:76–77. 40. Boysen M, Loven JO. Second malignant neoplasms in patients with head and neck squamous cell carcinomas. Acta Oncol 1993;32:283–288. 41. Braun IF, Hoffman JC Jr. Computed tomography of the buccomasseteric region. 1. Anatomy. AJNR Am J Neuroradiol 1984;5:605–610. 42. Braun IF, Hoffman JC Jr, Reede D, et al. Computed tomography of the buccomasseteric region. 2. Pathology. AJNR Am J Neuroradiol 1984;5:611–616. 43. Brazeau-Lamontagne L, Charlin B, Levesque R-Y, et al. Cricoarytenoiditis: CT assessment in rheumatoid arthritis. Radiology l986;158:463–466. 44. Brennan SC, Redd WH, Jacobsen PB, et al. Anxiety and panic during magnetic resonance scans. Lancet 1988;2:512. 45. Briggs RJS, Gallimore AP, Phelps PD, et al. Laryngeal imaging by computerized tomography and magnetic resonance following radiation therapy: a need for caution. J Laryngol Otol 1993;107: 565–568. 46. Brown J, Wippold FJ 2nd. Practical MRI: a teaching file. New York: Lippincott-Raven Press, 1996:200–220. 47. Capitanio MA, Kirkpatrick JA Jr. Upper respiratory tract obstruction in infants and children. Radiol Clin North Am 1968;6: 265–277. 48. Carbone A, Vaccher E, Barzan L, et al. Head and neck lymphomas associated with human immunodeficiency virus infection. Arch Otolaryngol Head Neck Surg 1995;121:210–218. 49. Carvalho P, Baldwin D, Carter R, et al. Accuracy of CT in detecting squamous carcinoma metastases in cervical lymph nodes. Clin Radiol 1991;44:79–81. 50. Casselman JW, Biebau G. Imaging of laryngeal cancer. Acta Otorhinolaryngol Belg 1992;46:161–174. 51. Castelijns JA. Diagnostic radiology in head and neck oncology. Curr Opin Oncol 1990;2:557–561. 52. Castelijns JA. Diagnostic radiology of head and neck oncology. Curr Opin Oncol 1991;3:512–518. 53. Castelijns JA, Doornbos J, Verbeeten B Jr, et al. MR imaging of the normal larynx. J Comput Assist Tomogr 1985;9:9l9–925. 54. Castelijns JA, Gerritsen GJ, Kaiser MC, et al. Invasion of laryngeal cartilage by cancer: comparison of CT and MR imaging. Radiology 1988;167:199–206. 55. Castelijns JA, Gerritsen GJ, Valk J, et al. MRI of normal or cancerous laryngeal cartilages: histopathologic correlation. Laryngoscope 1987;97:1085–1093. 56. Castelijns JA, Kaiser MC, Valk J, et al. MR imaging of laryngeal cancer. J Comput Assist Tomogr 1987;11:134–140. 57. Castelijns JA, van den Brekel MWM, Niekoop VA, et al. Imaging of the larynx. Neuroimaging Clin N Am 1996;6:401–415. 58. Castelijns JA, van den Brekel MWM, Tobi H, et al. Laryngeal carcinoma after radiation therapy: correlation of abnormal MR imaging signal patterns in laryngeal cartilage with the risk of recurrence. Radiology 1996;198:151–155. 59. Chakeres DW, Zawodniak LJ, Bornstein RA, et al. MR of head and neck adenopathy in asymptomatic HIV-seropositive men. AJNR Am J Neuroradiol 1993;14:1367–1371. 60. Charkes ND, Maurer AH, Siegel JA, et al. MR imaging in thyroid disorders: correlation of signal intensity with Graves disease activity. Radiology 1987;164:491–494. 61. Charlin B, Brazeau-Lamontagne L, Guerrier B, et al. Assessment of laryngeal cancer: CT scan versus endoscopy. J Otolaryngol 1989;18:283–288.

62. Charnoff SK, Carter BL. Plunging ranula: CT diagnosis. Radiology 1986;158: 467–468. 63. Chong J, Hinckley LK, Ginsberg LE. Masticator space abnormalities associated with mandibular osteoradionecrosis: MR and CT findings in five patients. AJNR Am J Neuroradiol 2000;21:175–178. 64. Chong VFH, Fan Y-F, Mukherji SK. Carcinoma of the nasopharynx. Semin Ultrasound CT MR 1998;19:449–462. 65. Chung CJ, Fordham LA, Mukherji SK. The pediatric airway: a review of differential diagnosis by anatomy and physiology. Neuroimaging Clin N Am 2000;10:161–180. 66. Chung TS, Yousem DM, Seigerman HM, et al. MR of mandibular invasion in patients with oral and oropharyngeal malignant neoplasms. AJNR Am J Neuroradiol 1994;15:1949–1955. 67. Coit WE, Harnsberger HR, Osborn AG, et al. Ranulas and their mimics: CT evaluation. Radiology 1987;163: 211–216. 68. Cunningham MJ. The management of congenital neck masses. Am J Otolaryngol 1992;13:78–92. 69. Curtin HD. Imaging of the larynx: current concepts. Radiology 1989;173:1–11. 70. Curtin HD. Separation of the masticator space from the parapharyngeal space. Radiology 1987;163: 195–204. 71. Curtin HD. The larynx. In: Som PM, Curtin HD, eds. Head and neck imaging, 3rd ed. St. Louis: Mosby Year Book, 1996:612–707. 72. Das Gupta TK, Brasfield RD, Strong EW, et al. Benign solitary schwannomas (neurilemomas). Cancer 1969;24:355–366. 73. Davis WL, Harnsberger HR, Smoker WRK, et al. Retropharyngeal space: evaluation of normal anatomy and diseases with CT and MR imaging. Radiology 1990;174:59–64. 74. Davis WL, Smoker WRK, Harnsberger HR. The normal and diseased retropharyngeal and prevertebral spaces. Semin Ultrasound CT MR 1990;11:520–523. 75. de Kerviler E, Bely N, Laccourreye O, et al. The aryepiglottic fold as a rare location of adenoid cystic carcinoma. AJNR Am J Neuroradiol 1995;16:1375–1377. 76. Deschler DG, Hayden RE. Head and neck reconstruction. Neuroimaging Clin N Am 1996;6:505–514. 77. de Serres LM, Sie KCY, Richardson MA. Lymphatic malformations of the head and neck. A proposal for staging. Arch Otolaryngol Head Neck Surg 1995;121:577–582. 78. Devaney KO, Ferlito A, Silver CE. Cartilaginous tumors of the larynx. Ann Otol Rhinol Laryngol 1995;104:251–255. 79. Dietrich RB, Lufkin RB, Kangarloo H, et al. Head and neck MR imaging in the pediatric patient. Radiology 1986;159:769–776. 80. Dillon WP. Applications of magnetic resonance imaging to the head and neck. Semin Ultrasound CT MR 1986;7:202–215. 81. Dillon WP, Harnsberger HR. The impact of radiologic imaging on staging of cancer of the head and neck. Semin Oncol 1991;18:64–79. 82. Dillon WP, Mills CM, Kjos B, et al. Magnetic resonance imaging of the nasopharynx. Radiology 1984;152:731–738. 83. DiSantis DJ, Balfe DM, Hayden RE, et al. The neck after total laryngectomy: CT study. Radiology 1984;153:713–717. 84. DiSantis DJ, Balfe DM, Hayden R, et al. The neck after vertical hemilaryngectomy: computed tomographic study. Radiology 1984;151:683–687. 85. Doppman JL, Krudy AG, Marx SJ, et al. Aspiration of enlarged parathyroid glands for parathyroid hormone assay. Radiology 1983;148:31–35. 86. Doppman JL, Shawker TH, Krudy AG, et al. Parathymic parathyroid: CT, US, and angiographic findings. Radiology 1985;157:419–423. 87. Duda JJ Jr, Lewin JS, Eliachar I. MR evaluation of epiglottic disruption. AJNR Am J Neuroradiol 1996;17:563–566. 88. Dujovny M, Kossovsky N, Kossowsky R, et al. Aneurysm clip motion during magnetic resonance imaging: in vivo experimental study with metallurgical factor analysis. Neurosurgery 1985;17: 543–548. 89. Dullerud R, Johansen JG, Dahl T, et al. Influence of CT on tumor classification of laryngeal carcinomas. Acta Radiol 1992;33:314–318. 90. Dunbar JS. Epiglottitis and croup. J Can Assoc Radiol 1961;12:86–95. 91. Dunbar JS. Upper respiratory tract obstruction in infants and children. Am J Roentgenol Radium Ther Nucl Med 1970;109: 227–246. 92. Dunn V, Coffman CE, McGowan JE, et al. Mechanical ventilation during magnetic resonance imaging. Magn Reson Imaging 1985;3:169–172.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 217

Neck 93. Edelman RR, Shellock FG, Ahladis J. Practical MRI for the technologist and imaging specialist. In: Edelman RR, Hesselink JR, eds. Clinical magnetic resonance imaging. Philadelphia: Saunders, 1990:39–72. 94. Elango S. Parapharyngeal space lipoma. Ear Nose Throat J 1995; 74:52–53. 95. Emami B, Purdy J, Graham M, et al. Clinical Ie with 3-D radiotherapy: the Washington University Ie. In: Purdy JA, Emami B, eds. 3-D radiation treatment planning and conformal therapy: proceedings of an international symposium. Madison, WI: Medical Physics Publishing, 1995:121–124. 96. Emami B, Purdy JA, Simpson JR, et al. 3-D conformal radiotherapy in head and neck cancer. The Washington University experience. Front Radiat Ther Oncol 1996;29:207–220. 97. Escott EJ. A variety of appearances of malignant melanoma in the head: a review. Radiographics 2001;21:625–639. 98. Fabrikant H. Parapharyngeal extension of a meningioma. Australas Radiol 1993;37:102–104. 99. Ferlito A. Secondary tumors of the larynx. In: Ferlito A. Cancer of the larynx, vol. 2. Boca Raton, FL: CRC Press, 1985:89–108. 100. Ferlito A, Barnes L, Wenig BM. Identification, classification, treatment, and prognosis of laryngeal paraganglioma: review of the literature and eight new cases. Ann Otol Rhinol Laryngol 1994; 103:525–536. 101. Ferlito A, Milroy CM, Wenig BM, et al. Laryngeal paraganglioma versus atypical carcinoid tumor. Ann Otol Rhinol Laryngol 1995; 104:78–83. 102. Finn EJ, Di Chiro G, Brooks RA, et al. Ferromagnetic material in patients: detection before MR imaging. Radiology 1985;156: 139–141. 103. Fishbain DA, Goldberg M, Labbé E, et al. Long-term claustrophobic following magnetic resonance imaging. Am J Psychiatry 1988;145:1038–1039. 104. Fordham LA, Chung CJ, Donnelly LF. Imaging of congenital vascular and lymphatic anomalies of the head and neck. Neuroimaging Clin N Am 2000;10:117–136. 105. Freitas JE, Freitas AE. Thyroid and parathyroid imaging. Semin Nucl Med 1994;24:234–245. 106. Friday PJ, Kubal WS. Magnetic resonance imaging: improved patient tolerance utilizing medical hypnosis. Am J Clin Hypn 1990;33:80–84. 107. Friday PJ, Westman HR. Reducing anxiety in patients undergoing MRI. MRI Decis 1991;5:29–31. 108. Friedman L, Patel M, Steinberg J, et al. CT appearance of an oncocytic papillary cystadenoma of the larynx. J Comput Assist Tomogr 1990;14:322–324. 109. Friedman M, Roberts N, Kirshenbaum GL, et al. Nodal size of metastatic squamous cell carcinoma of the neck. Laryngoscope 1993;103: 854–856. 110. Friedman M, Shelton VK, Mafee M, et al. Metastatic neck disease: evaluation by computed tomography. Arch Otolaryngol 1984; 110:443–447. 111. Fruin ME, Smoker WRK, Harnsberger HR. The carotid space in the suprahyoid neck. Semin Ultrasound CT MR 1990;11:504–519. 112. Fulbright R, Panush D, Sze G, et al. MR of the head and neck: comparison of fast spin-echo and conventional spin-echo sequences. AJNR Am J Neuroradiol 1994;15:767–773. 113. Funari M, Campos Z, Gooding GAW, et al. MRI and ultrasound detection of asymptomatic thyroid nodules in hyperparathyroidism. J Comput Assist Tomogr 1992;16:615–619. 114. Gamsu G. Computed tomography of the larynx and piriform sinuses. In: Moss AA, Gamsu G, Genant H, eds. Computed tomography of the body. Philadelphia: WB Saunders, 1983:65–143. 115. Gamsu G, Mark AS, Webb WR. Computed tomography of the normal larynx during quiet breathing and phonation. J Comput Assist Tomogr 1981;5:353–360. 116. Gamsu G, Webb WR, Shallit JB, et al. CT in carcinoma of the larynx and pyriform sinus: value of phonation scans. AJR Am J Roentgenol 1981;136:577–584. 117. Gao P-Y, Osborn AG, Smirniotopoulos JG, et al. Radiologicpathologic correlation.Epidermoid tumor of the cerebellopontine angle. AJNR Am J Neuroradiol 1992:13:863–872. 118. Gatenby RA, Mulhern CB Jr, Richter MP, et al. CT-guided biopsy for the detection and staging of tumors of the head and neck. AJNR Am J Neuroradiol 1984;5:287–289.

217

119. Gatenby RA, Mulhern CB Jr, Strawitz J. CT-guided percutaneous biopsies of head and neck masses. Radiology 1983;146:717–719. 120. Gatenby RA, Mulhern CB Jr, Strawitz J, et al. Comparison of clinical and computed tomographic staging of head and neck tumors. AJNR Am J Neuroradiol 1985;6:399–401. 121. Gayler BW, Kashima HK, Martinez CR. Computed tomography of the neck. Crit Rev Diagn Imaging 1985;23:319–376. 122. Gefter WB, Spritzer CE, Eisenberg B, et al. Thyroid imaging with high-field-strength surface-coil MR. Radiology 1987;l64:483–490. 123. George CD, Ng YY, Hall-Craggs MA, et al. Parotid haemangioma in infants: MR imaging at 1.5T. Pediatr Radiol 1991;21:483–485. 124. Gillams AR, Fuleihan N, Grillone G, et al. Magnetization transfer contrast MR in lesions of the head and neck. AJNR Am J Neuroradiol 1996;17:355–360. 125. Giron J, Joffre P, Serres-Cousine O, et al. Pre-therapeutic evaluation of laryngeal carcinomas using computed tomography and magnetic resonance imaging. Isr J Med Sci 1992;28:225–232. 126. Giron J, Joffre P, Serres-Cousine O, et al. CT and MR evaluation of laryngeal carcinomas. J Otolaryngol 1993;22:284–293. 127. Glasier CM, Stark JE, Jacobs RF, et al. CT and ultrasound imaging of retropharyngeal abscesses in children. AJNR Am J Neuroradiol 1992;13:1191–1195. 128. Glastonbury CM, Davidson HC, Haller JR, et al. The CT and MR imaging features of carcinoma arising in thyroglossal duct remnants. AJNR Am J Neuroradiol 2000;21:770–774. 129. Glazer HS, Aronberg DJ, Lee JKT, et al. Extralaryngeal causes of vocal cord paralysis: CT evaluation. AJR Am J Roentgenol 1983; 141:527–531. 130. Glazer HS, Lee JKT, Levitt RG, et al. Radiation fibrosis: differentiation from recurrent tumor by MR imaging. Radiology 1985;156: 721–726. 131. Glazer HS, Mauro MA, Aronberg DJ, et al. Computed tomography of laryngocoeles. AJR Am J Roentgenol 1983;140:549–552. 132. Glazer HS, Niemeyer JH, Balfe DM, et al. Neck neoplasms: MR imaging. Part I. Initial evaluation. Radiology 1986;160: 343–348. 133. Glazer HS, Niemeyer JH, Balfe DM, et al. Neck neoplasms: MR imaging. Part II. Posttreatment evaluation. Radiology 1986; 160: 349–354. 134. Glazer HS, Sagel SS. The larynx and hypopharynx. In: Carter BL, ed. Computed tomography of the head and neck. New York: Churchill Livingstone, 1985:1–30. 135. Gold BM. Second branchial cleft cyst and fistula. AJR Am J Roentgenol 1980;134:1067–1069. 136. Graney DO, Flint PW. Anatomy. In: Cummings CW, Frederickson JM, Harker LA, et al., eds. Otolaryngology head and neck surgery. St. Louis: Mosby Year Book, 1993:1693–1703. 137. Greene FL, Page DL, Fleming ID, et al. AJCC cancer staging manual, 6th ed. New York: Springer, 2002. 138. Gregor RT, Loftus B, Cohen P, et al. Saccular mucocele in association with laryngeal cancer. Ann Otol Rhinol Laryngol 1994;103: 732–736. 139. Griffin JL, Ramadan HH, Wetmore SJ. Laryngocele: a cause of stridor and airway obstruction. Otolaryngol Head Neck Surg 1993;108:760–762. 140. Gritzmann N, Grasi MC, Helmer M, et al. Invasion of the carotid artery and jugular vein by lymph node metastases: detection with sonography AJR Am J Roentgenol 1990;154:411– 414. 141. Gritzmann N, Traxler M, Grasl M, et al. Advanced laryngeal cancer: sonographic assessment. Radiology 1989;171:171–175. 142. Grodinsky M, Holyoke EA. The fasciae and fascial spaces of the head, neck and adjacent regions. Am J Anat 1938;63:367–407. 143. Guneri A, Ceryan K, Igci E, et al. View from within: radiology in focus. Lingual thyroid: the diagnostic value of magnetic resonance imaging. J Laryngol Otol 1991;105:493–495. 144. Hagiwara A, Inoue Y, Nakayama T, et al. The “botryoid sign”: a characteristic feature of rhabdomyosarcomas in the head and neck. Neuroradiology 2001;43:331–335. 145. Hardin CW, Harnsberger HR, Osborn AG, et al. Infection and tumor of the masticator space: CT evaluation. Radiology 1985;157:413–417. 146. Harkin JC, Reed RJ. Tumors of peripheral nervous system, 2nd series, fascicle 3. Washington, DC: Armed Forces Institute of Pathology, 1969:29–120.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 218

218

Chapter 4

147. Harnsberger HR. The infrahyoid neck: normal anatomy and pathology of the head and neck from the hyoid bone to the clavicles. In: Harnsberger HR. Handbook of head and neck imaging, 2nd ed. St. Louis: Mosby Year Book, 1995:150–198. 148. Harnsberger HR. The larynx and hypopharynx. In: Harnsberger HR. Handbook of head and neck imaging, 2nd ed. St. Louis: Mosby Year Book, 1995:224–260. 149. Harnsberger HR. The oral cavity: emphasizing the sublingual and submandibular spaces. In: Harnsberger HR. Head and neck imaging, 2nd ed. St. Louis: Mosby Year Book, 1995:120–149. 150. Harnsberger HR. The parotid space. In: Harnsberger HR. Head and neck imaging, 2nd ed. St. Louis: Mosby Year Book, 1995:60–74. 151. Harnsberger HR. The perivertebral space. In: Harnsberger HR. Head and neck imaging, 2nd ed. St. Louis: Mosby Year Book, 1995:105–119. 152. Harnsberger HR. The retropharyngeal space. In: Harnsberger HR. Head and neck imaging, 2nd ed. St. Louis: Mosby Year Book, 1995:89–104. 153. Harnsberger HR. Squamous cell carcinoma: nodal staging. In: Harnsberger HR. Handbook of head and neck imaging, 2nd ed St. Louis: Mosby, 1995:283–298. 154. Harnsberger HR, Dillon WP. The radiologic role in diagnosis, staging, and follow-up of neoplasia of the brain, spine, and head and neck. Semin Ultrasound CT MR 1989, 10:431–452. 155. Harnsberger HR, Mancuso AA, Muraki AS, et al. Branchial cleft anomalies and their mimics: computed tomographic evaluation. Radiology 1984;152:739–748. 156. Harnsberger HR, Mancuso AA, Muraki AS, et al. The upper aerodigestive tract and neck: CT evaluation of recurrent tumors. Radiology 1983;149:503–509. 157. Harnsberger HR, Osborn AG. Differential diagnosis of head and neck lesions based on their space of origin. 1. The suprahyoid part of the neck. AJR Am J Roentgenol 1991;157:147–154. 158. Harris KG, Smith TP, Cragg AH, et al. Nephrotoxicity from contrast material in renal insufficiency: ionic versus nonionic agents. Radiology 1991;179:849–852. 159. Hasegawa Y, Matsuura H. Retropharyngeal node dissection in cancer of the oropharynx and hypopharynx. Head Neck 1994; 16:173–180. 160. Hasso AN, Brown KD. Use of gadolinium chelates in MR imaging of lesions of the extracranial head and neck. J Magn Reson Imaging 1993;3:247–263. 161. Haustein J, Niendorf HP, Krestin G, et al. Renal tolerance of gadolinium-DTPA/dimeglumine in patients with chronic renal failure. Invest Radiol 1992;27:153–156. 162. Hayman LA, Taber KH, Diaz-Marchan PJ, et al. Spatial compartments of the neck, part I: coronal sections. Int J Neuroradiol 1998;4:233–239. 163. Hermans R, Pameijer FA, Mancuso AA, et al. Laryngeal or hypopharyngeal squamous cell carcinoma: can follow-up CT after definitive radiation therapy be used to detect local failure earlier than clinical examination alone? Radiology 2000;214:683–687. 164. Hermans R, Van den Bogaert W, Rijnders A, et al. Value of computed tomography as outcome predictor of supraglottic squamous cell carcinoma treated by definitive radiation therapy. Int J Radiat Oncol Biol Phys 1999;44:755–765. 165. Hiatt JL, Gartner LP. Palate, pharynx and larynx. In: Hiatt JL, Gartner LP. Text of head and neck anatomy. Baltimore: Williams and Wilkins, 1987:245–264. 166. Hirano M, Kurita S, Cho JS, et al. Computed tomography in determining laryngeal involvement of hypopharyngeal carcinoma. Ann Otol Rhinol Laryngol 1988;97:476–482. 167. Hoe J. CT of nasopharyngeal carcinoma: significance of widening of the preoccipital soft tissue on axial scans. AJR Am J Roentgenol 1989;153:867–872. 168. Hoover LA, Calcaterra TC, Walter GA, et al. Preoperative CT scan evaluation for laryngeal carcinoma: correlation with pathologic findings. Laryngoscope 1984;94:310–315. 169. Hoover LA, Wortham DG, Lufkin RB, et al. Magnetic resonance imaging of the larynx and tongue base: clinical applications. Otolaryngol Head Neck Surg 1987;97:245–256. 170. Hopper KD. Questions and answers. AJR Am J Roentgenol 1996; 166:715. 171. Hudgins PA. Nodal and nonnodal inflammatory processes of the pediatric neck. Neuroimaging Clin N Am 2000;10:181–192.

172. Hudgins PA. The nasopharynx and related spaces. In: American society of head and neck radiology syllabus, 26th Annual Conference and Post Graduate Course, Vancouver, BC; Chicago: ASHNR;1993:73–80. 173. Hudgins PA, Burson JG, Gussack GS et al. CT and MR appearance of recurrent malignant head and neck neoplasms after resection and flap reconstruction. AJNR Am J Neuroradiol 1994;15: 1689–1694. 174. Hudgins PA, Elster AD, Runge VM, et al. Efficacy and safety of gadopentetate dimeglumine in the evaluation of patients with a suspected tumor of the extracranial head and neck. J Magn Reson Imaging 1993;3:345–349. 175. Hudgins PA, Gussack GS. MR imaging in the management of extracranial malignant tumors of the head and neck. AJR Am J Roentgenol 1992;159:161–169. 176. Hurst RW. Interventional neuroradiology of the head and neck. Neuroimaging Clin N Am 1996;6:473–495. 177. Hurtado JF, Lopez JJ, Aranda FI, et al. Primary liposarcoma of the larynx: case report and literature review. Ann Otol Rhinol Laryngol 1994;103:315–318. s 179. Jabour BA, Lufkin R, Hanafee W. The larynx and trachea. Curr Opin Radiol 1990;2:118–121. 180. Jabour BA, Lufkin RB, Hanafee WN. Magnetic resonance imaging of the larynx. Top Magn Reson Imaging 1990;2:60–68. 181. Jabour BA, Lufkin RB, Layfield LJ, et al. Magnetic resonance imaging of metastatic cervical adenopathy. Top Magn Reson Imaging 1990;2:69–75. 182. Jacobs CJM, Harnsberger HR, Lufkin RB, et al. Vagal neuropathy: evaluation with CT and MR imaging. Radiology 1987;164:97–102. 183. Jinkins JR. Computed tomography of the cranio-cervical lymphatic system: anatomical and functional considerations. Neuroradiology 1987;29:317–326. 184. Johnson JT. A surgeon looks at cervical lymph nodes. Radiology 1990;175:607–610. 185. Johnston EF, Hammond AJ, Cairncross JG. Bilateral hypoglossal palsies: a late complication of curative radiotherapy. Can J Neurol Sci 1989;16:198–199. 186. Jungehuelsing M, Sittel C, Fischbach R, et al. Limitations of magnetic resonance imaging in the evaluation of perineural tumor spread causing facial nerve paralysis. Arch Otolaryngol Head Neck Surg 2000;126:506–510. 187. Keberle M, Kenn W, Hahn D. Current concepts in imaging of laryngeal and hypopharyngeal cancer. Eur Radiol 2002;12:1672–1683. 188. Kahn JL, Wolfram-Gabel R, Bourjat P. Anatomy and imaging of the deep fat of the face. Clin Anat 2000;13:373–382. 189. Kaji AV, Mohuchy T, Swartz JD. Imaging of cervical lymphadenopathy. Semin Ultrasound CT MR 1997;18:220–249. 190. Kalender WA, Vock P, Polacin A, et al. Spiral CT: a new technique for volumetric scans. I. Basic principles and methodology.] Röntgenpraxis 1990;43:323–330. German. 191. Kallmes DF, Phillips CD. The normal anterior commissure of the glottis. AJR Am J Roentgenol 1997;168:1317–1319. 192. Kansal P, Sakati N, Rifai A, et al. Lingual thyroid: diagnosis and treatment. Arch Intern Med 1987;147:2046–2048. 193. Kelly WM, Paglen PG, Pearson JA, et al. Ferromagnetism of intraocular foreign body causes unilateral blindness after MR study. AJNR Am J Neuroradiol 1986;7:243–245. 194. Kennedy KS, Parker GS. Subglottic and tracheal malignancies. Ear Nose Throat J 1992;71:233–237. 195. Kier R, Blinder RA, Herfkens RJ, et al. MR imaging with surface coils in primary hyperparathyroidism. J Comput Assist Tomogr 1987;11:863–868. 196. Kim YJ, Sung KJ, Kim MS, et al. CT manifestations of cervical tuberculous lymphadenitis. J Otolaryngol 1993;22:321–325. 197. Kirchner JA. Two hundred laryngeal cancers: patterns of growth and spread as seen in serial section. Laryngoscope 1977;87:474–482. 198. Kirshenbaum KJ, Nadimpalli SR, Friedman M, et al. Benign lymphoepithelial parotid tumors in AIDS patients: CT and MR findings in nine cases. AJNR Am J Neuroradiol 1991;12: 271–274. 199. Klucznik RP, Carrier DA, Pyka R, et al. Placement of a ferromagnetic intracerebral aneurysm clip in a magnetic field with a fatal outcome. Radiology 1993;187:855–856.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 219

Neck 200. Kneeland JB, Krubsack AJ, Lawson TL, et al. Enlarged parathyroid glands: high-resolution local coil MR imaging. Radiology 1987; 162:143–146. 201. Koch WM, Patel H, Brennan J, et al. Squamous cell carcinoma of the head and neck in the elderly. Arch Otolaryngol Head Neck Surg 1995;121:262–265. 202. Koeller KK, Alamo L, Adair CF, et al. Congenital cystic masses of the neck: radiologic-pathologic correlation. Radiographics 1999; 19:121–146. 203. Kolbenstvedt A, Charania B, Natvig K, et al. Computed tomography in T1 carcinoma of the larynx. Acta Radiol 1989;30:467–469. 204. Kraus M, Fliss DM, Leiberman A, et al. Mycobacterial cervical adenitis. Ann Otol Rhinol Laryngol 1995;104:409–412. 205. Krinsky G, Rofsky NM, Weinreb JC. Nonspecificity of short inversion time inversion recovery (STIR) as a technique of fat suppression: pitfalls in image interpretation. AJR Am J Roentgenol 1996;166:523–526. 206. Kurabayashi T, Ida M, Yasumoto M, et al. MRI of ranulas. Neuroradiology 2000;42:917–922. 207. Kurita S, Hirano M, Matsuoka H, et al. A histopathological study of carcinoma of the larynx. Auris Nasus Larynx 1985;12:S172– S177. 208. Kushner DC, Harris GBC. Obstructing lesions of the larynx and trachea in infants and children. Radiol Clin North Am 1978;16: 181–194. 209. Kwock L. Localized MR spectroscopy: basic principles. Neuroimaging Clin N Am 1998;8:713–731. 210. Laine FJ, Braun IF, Jensen ME, et al. Perineural tumor extension through the foramen ovale: evaluation with MR imaging. Radiology 1990;174:65–71. 211. Lanzieri CF, Bangert B. Magnetic resonance imaging of the nasopharynx. Top Magn Reson Imaging 1990;2:39–47. 212. Lakshminarayanan AV, Lee S, McCutcheon MJ. MR imaging of the vocal tract during vowel production. J Magn Reson Imaging 1991;1:71–76. 213. Lambre H, Anzai Y, Farahani K, et al. Interventional magnetic resonance imaging of the head and neck and new imaging techniques. Neuroimaging Clin N Am 1996;6:461–472. 214. Larsson SG, Mancuso A, Hanafee W. Computed tomography of the tongue and floor of the mouth. Radiology 1982;143: 493– 500. 215. Larsson S, Mancuso A, Hoover L, et al. Differentiation of pyriform sinus cancer from supraglottic laryngeal cancer by computed tomography. Radiology 1981;141:427–432. 216. Lewin JS, Curtin HD, Ross JS, et al. Fast spin-echo imaging of the neck: comparison with conventional spin-echo, utility of fat suppression, and evaluation of tissue contrast characteristics. AJNR Am J Neuroradiol 1994;15:1351–1357. 217. Loevner LA, Yousem DM, Montone KT, et al. Can radiologists accurately predict preepiglottic space invasion with MR imaging? AJR Am J Roentgenol 1997;169:1681–1687. 218. Lofchy NM, Stevens JK, Brown DH. Three-dimensional imaging of the parapharyngeal space. Arch Otolaryngol Head Neck Surg 1994;120:333–336. 219. Logue JP, Hale RJ, Stewart AL, et al. Primary malignant lymphoma of the thyroid: a clinicopathological analysis. Int J Radiat Oncol Biol Phys 1992;22:929–933. 220. Lowe LH, Stokes LS, Johnson JE, et al. Swelling at the angle of the mandible: imaging of the pediatric parotid gland and periparotid region. Radiographics 2001;21:1211–1227. 221. Lufkin R, Hanafee W. Application of surface coils to MRI anatomy of the larynx. AJR Am J Roentgenol 1985;145:483–489. 222. Lufkin RB, Hanafee WN, Wortham D, et al. Larynx and hypopharynx: MR imaging with surface coils. Radiology 1986;158: 747–754. 223. Lufkin RB, Larsson SG, Hanafee WN. Work in progress: NMR anatomy of the larynx and tongue base. Radiology 1983;148: 173–175. 224. Madison MT, Remley KB, Latchaw RE, et al. Radiologic diagnosis and staging of head and neck squamous cell carcinoma. Radiol Clin North Am 1994;32:163–181. 225. Mafee MF, Schild JA, Valvassori GE, et al. Computed tomography of the larynx: correlation with anatomic and pathologic studies in cases of laryngeal carcinoma. Radiology 1983; 147:123–128.

219

226. Mancuso AA. Cervical lymph node metastases. In: Bragg DG, Rubin P, Youker JE, eds. Oncologic imaging. New York: Pergamon Press, 1985:47–61. 227. Mancuso AA. Larynx and hypopharynx. In: Bragg DG, Rubin P, Youker JE, eds. Oncologic imaging. New York: Pergamon Press, 1985:103–123. 228. Mancuso AA, Bohman L, Hanafee W, et al. Computed tomography of the nasopharynx: normal and variants of normal. Radiology 1980;137:113–121. 229. Mancuso AA, Calcaterra TC, Hanafee WN. Computed tomography of the larynx. Radiol Clin North Am 1978;16:195–208. 230. Mancuso AA, Hanafee WN. A comparative evaluation of computed tomography and laryngography. Radiology 1979;133: 131–138. 231. Mancuso AA, Hanafee WN. Computed tomography and magnetic resonance imaging of the head and neck, 2nd ed. Baltimore: Williams & Wilkins, l985. 232. Mancuso AA, Hanafee WN. Computed tomography of the injured larynx. Radiology 1979;133:139–144. 233. Mancuso AA, Harnsberger HR, Muraki AS, et al. Computed tomography of cervical and retropharyngeal lymph nodes: normal anatomy, variants of normal, and applications in staging head and neck cancer. Part I: normal anatomy. Radiology 1983; 148:709–714. 234. Mancuso AA, Harnsberger HR, Muraki AS, et al. Computed tomography of cervical and retropharyngeal lymph nodes: normal anatomy, variants of normal, and applications in staging head and neck cancer. Part II: pathology. Radiology 1983;148:7l5–723. 235. Mancuso AA, Mukherji SK, Schmalfuss I, et al. Preradiotherapy computed tomography as a predictor of local control in supraglottic carcinoma. J Clin Oncol 1999;17:631–637. 236. Mancuso AA, Maceri D, Rice D, et al. CT of cervical lymph node cancer. AJR Am J Roentgenol 1981;136:381–385. 237. Mancuso AA, Tamakawa Y, Hanafee WN. CT of the fixed vocal cord. AJR Am J Roentgenol 1980;135:7529–7534. 238. Marks JE. Radiation therapy. In: Cummings CW, Frederickson JM, Harker LA, et al., eds. Otolaryngology–head and neck surgery. St. Louis: Mosby Year Book, 1993:2122–2147. 239. Martinez CR, Gayler B, Kashima H, et al. Computed tomography of the neck. Radiographics 1983;3:9–43. 240. Matsumoto AH, Teitelbaum GP, Barth KH, et al. Tantalum vascular stents: in vivo evaluation with MR imaging. Radiology 1989; 170:753–755. 241. McClennan BL. Low-osmolality contrast media: premises and promises. Radiology 1987;162:1–8. 242. McCook TA, Kirks DR. Epiglottic enlargement in infants and children. Another radiologic look. Pediatr Radiol 1982;12: 227–234. 243. McDermott VG, Fernandez RJM, Meakem TJ 3rd, et al. Preoperative MR imaging in hyperparathyroidism: results and factors affecting parathyroid detection. AJR Am J Roentgenol 1996;166: 705–710. 244. McGuirt WF Jr, Little JP. Laryngeal cancer in children and adolescents. Otolaryngol Clin North Am 1997;30:207–214. 245. McKenna KM, Jabour BA, Lufkin RB, et al. Magnetic resonance imaging of the tongue and oropharynx. Top Magn Reson Imaging 1990;2:49–59. 246. McManus K, Holt GR, Aufdemorte TM, et al. Bronchogenic cyst presenting as deep neck abscess. Otolaryngol Head Neck Surg 1984;92:109–114. 247. Merritt RM, Williams MF, James TH, et al. Detection of cervical metastasis: a meta-analysis comparing computed tomography with physical examination. Arch Otolaryngol Head Neck Surg 1997; 123:149–152. 248. Miller MB, Rao VM, Tom BM. Cystic masses of the head and neck: pitfalls in CT and MR interpretation. AJR Am J Roentgenol 1992;159:601–607. 249. Million RR. The larynx... so to speak: everything I wanted to know about laryngeal cancer I learned in the last 32 years. Int J Radiat Oncol Biol Phys 1992;23:691–704. 250. Minami M, Tanioka H, Oyama K, et al. Warthin tumor of the parotid gland: MR-pathologic correlation. AJNR Am J Neuroradiol 1993;14:209–214. 251. Mokri B, Schievink WI, Olsen KD, et al. Spontaneous dissection of the cervical internal carotid artery: presentation with lower cranial nerve palsies. Arch Otolaryngol Head Neck Surg 1992; 118:431–435.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 220

220

Chapter 4

252. Molony TB, Brackmann DE, Lo WWM. Meningiomas of the jugular foramen. Otolaryngol Head Neck Surg 1992;106:128–136. 253. Moon WK, Han MH, Chang KH, et al. Laryngeal tuberculosis: CT findings. AJR Am J Roentgenol 1996;166:445–449. 254. Mukherji SK, Castelijns J, Castillo M. Squamous cell carcinoma of the oropharynx and oral cavity: how imaging makes a difference. Semin Ultrasound CT MR 1998;19:463–475. 255. Mukherji SK, Castillo M. A simplified approach to the spaces of the suprahyoid neck. Radiol Clin North Am 1998;36:761–780. 256. Mukherji SK, Castillo M, Huda W, et al. Comparison of dynamic and spiral CT for imaging the glottic larynx. J Comput Assist Tomogr 1995;19:899–904. 257. Mukherji SK, Chong V, Castillo M. Magnetic resonance spectroscopy of the extracranial head and neck. In: Mukherji SK, Castelijns JA, eds. Modern head and neck imaging. Berlin: SpringerVerlag, 2000;31–48. 258. Mukherji SK, Drane WE, Mancuso AA, et al. Occult primary tumors of the head and neck: detection with 2-[F-18] fluoro-2deoxy-D-glucose SPECT. Radiology 1996;199:761–766. 259. Mukherji SK, Fatterpekar G, Castillo M, et al. Imaging of congenital anomalies of the branchial apparatus. Neuroimaging Clin N Am 2000;10:75–93. 260. Mukherji SK, Kasper ME, Tart RP, et al. Irradiation paragangliomas of the head and neck: CT and MR appearance. AJNR Am J Neuroradiol 1994;15:357–363. 261. Mukherji SK, Mancuso AA, Kotzur IM, et al. Radiologic appearance of the irradiated larynx. Part II. Primary site response. Radiology 1994;193:149–154. 262. Mukherji SK, Mancuso AA, Kotzur IM, et al. Radiologic appearance of the irradiated larynx. Part I. Expected changes. Radiology 1994;193:141–148. 263. Mukherji SK, O’Brien SM, Gerstle RJ, et al. The ability of tumor volume to predict local control in surgically treated squamous cell carcinoma of the supraglottic larynx. Head Neck 2000; 22:282–287. 264. Mukherji SK, Schiro S, Castillo M. Proton MR spectroscopy of squamous cell carcinoma of the upper aerodigestive tract: in vitro characteristics. AJNR Am J Neuroradiol 1996;17:1485–1490. 265. Mukherji SK, Schiro S, Castillo M, et al. Proton MR spectroscopy of squamous cell carcinoma of the extracranial head and neck: in vitro and in vivo studies. AJNR Am J Neuroradiol 1997;18:1057–1072. 266. Mukherji SK, Tart RP, Slattery WH, et al. Evaluation of first branchial anomalies by CT and MR. J Comput Assist Tomogr 1993;17:576–581. 267. Mukherji SK, Turetsky D, Tart RP, et al. A technique for core biopsies of head and neck masses. AJNR Am J Neuroradiol 1994;15:518–520. 268. Muñoz A, Ramos A, Ferrando J, et al. Laryngeal carcinoma: sclerotic appearance of the cricoid and arytenoid cartilage—CTpathologic correlation. Radiology 1993;189:433–437. 269. Muraki AS, Mancuso AA, Harnsberger HR. Metastatic cervical adenopathy from tumors of unknown origin: the role of CT. Radiology 1984;152:749–753. 270. Muraki AS, Mancuso AA, Harnsberger HR, et al. CT of the oropharynx, tongue base, and floor of the mouth: normal anatomy and range of variations, and applications in staging carcinoma. Radiology 1983;148:725–731. 271. Nelson KL, Runge VM. Basic principles of MR contrast. Top Magnet Reson Imaging 1995;7:124–136. 272. New PFJ, Rosen BR, Brady TJ, et al. Potential hazards and artifacts of ferromagnetic and nonferromagnetic surgical and dental materials and devices in nuclear magnetic resonance imaging. Radiology 1983;147:139–148. 273. Niemeyer JH, Balfe DM, Hayden RE. Neck evaluation with barium-enhanced radiographs and CT scans after supraglottic subtotal laryngectomy. Radiology 1987;162:493–498. 274. Noma S, Nishimura K, Togashi K, et al. Thyroid gland: MR imaging. Radiology 1987;164:495–499. 275. Nömayr A, Lell M, Sweeney R, et al. MRI appearance of radiation-induced changes of normal cervical tissues. Eur Radiol 2001;11:1807–1817. 276. Ogura JH, Biller HF. Conservation surgery in cancer of the head and neck. Otolaryngol Clin North Am 1969;2:64l–665.

277. Ogura JH, Henneman H. Conservation surgery of the larynx and hypopharynx—selection of patients and results. Can J Otolaryngol 1973;2:11–16. 278. Ogura JH, Sessions DG, Spector GJ. Conservation surgery for epidermoid carcinoma of the supraglottic larynx. Laryngoscope 1975;85:1808–1815. 279. Ohgi K, Kohno A, Itabashi K, et al. Head and neck reconstruction with pectoralis major myocutaneous flap: CT evaluation. J Comput Assist Tomogr 1990;14:286–290. 280. Ohnishi T, Noguchi S, Murakami N, et al. Detection of recurrent thyroid cancer: MR versus thallium-201 scintigraphy. AJNR Am J Neuroradiol 1993;14:1051–1057. 281. Ohnishi T, Noguchi S, Murakami N, et al. MR imaging in patients with primary thyroid lymphoma. AJNR Am J Neuroradiol 1992;13:1196–1198. 282. Olmi P, Fallai C, Colagrande S, et al. Staging and follow-up of nasopharyngeal carcinoma: magnetic resonance imaging versus computerized tomography. Int J Radiat Oncol Biol Phys 1995;32:795–800. 283. Olsen WL, Dillon WP, Kelly WM, et al. MR imaging of paragangliomas. AJR Am J Roentgenol 1987;148:201–204. 284. Paff GH. Lymph nodes and lymphatics of head and neck. In: Paff GH. Anatomy of the head and neck. Philadelphia: Saunders, 1973:221–228. 285. Papazoglou G, Doundoulakis G, Terzakis G, et al. Pharyngocutaneous fistula after total laryngectomy: incidence, cause, and treatment. Ann Otol Rhinol Laryngol 1994;103:801–805. 286. Parker GD, Harnsberger HR, Jacobs JM. The pharyngeal mucosal space. Semin Ultrasound CT MR 1990;11:460–475. 287. Parker GD, Harnsberger HR. Radiologic evaluation of the normal and diseased posterior cervical space. AJR Am J Roentgenol 1991;157:161–165. 288. Pavlicek W, Geisinger M, Castle L, et al. The effects of nuclear magnetic resonance on patients with cardiac pacemakers. Radiology 1983;147:149–153. 289. Peck WW, Higgins CB, Fisher MR, et al. Hyperparathyroidism: comparison of MR imaging with radionuclide scanning. Radiology 1987;163:4l5–420. 290. Peel RZ, Gnepp DR. Diseases of the salivary glands. In: Barnes L, ed. Surgical pathology of the head and neck, vol. 1. New York: Marcel Dekker, 1985:533–645. 291. Phelps PD. Carcinoma of the larynx—the role of imaging in staging and pre-treatment assessments. Clin Radiol 1992;46: 77–83. 292. Plantet M-M, Hagay C, De Maulmont C, et al. Laryngeal schwannomas. Eur J Radiol 1995;21:61–66. 293. Pollei SR, Harnsberger HR. The radiologic evaluation of the parotid space. Semin Ultrasound CT MR 1990;11:486–503. 294. Price DB, Miller LJ, Price AP. Evaluation of nondistended pyriform sinus without respiratory maneuvers. J Comput Assist Tomogr 1993;17:163–165. 295. Price DC. Radioisotopic evaluation of the thyroid and the parathyroids. Radiol Clin North Am 1993;31:991–1015. 296. Quillin SP, Balfe DM, Glick SN. Pharyngography after head and neck irradiation: differentiation of postirradiation edema from recurrent tumor. AJR Am J Roentgenol 1993;161:1205–1208. 297. Quirk ME, Letendre AJ, Ciottone RA, et al. Evaluation of three psychologic interventions to reduce anxiety during MR imaging. Radiology 1989;173:759–762. 298. Quirk ME, Letendre AJ, Ciottone RA, et al. Anxiety in patients undergoing MR imaging. Radiology 1989;170:463–466. 299. Reede DL. Imaging modalities for the evaluation of neck pathology. Otolaryngol Clin North Am 1988;21:495–511. 300. Reede DL, Bergeron RT. The CT evaluation of the normal and diseased neck. Semin Ultrasound CT MR 1986;7:181–201. 301. Reede DL, Bergeron RT, McCauley DI. CT of the thyroid and of other thoracic inlet disorders. J Otolaryngol 1982;11:349–357. 302. Reede DL, Bergeron RT, Som PM. CT of thyroglossal duct cysts. Radiology 1985;157:121–125. 303. Reede DL, Bergeron RT, Whelan MA, et al. Computed tomography of cervical lymph nodes. Radiographics 1983;3:339–351. 304. Reede DL, Whelan MA, Bergeron RT. CT of the soft tissue structures of the neck. Radiol Clin North Am 1984;22:239–250.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 221

Neck 305. Reede DL, Whelan MA, Bergeron RT. Computed tomography of the infrahyoid neck. Part I: normal anatomy. Radiology 1982;145: 389–395. 306. Reede DL, Whelan MA, Bergeron RT. Computed tomography of the infrahyoid neck. Part II: pathology. Radiology 1982;145:397–402. 307. Reid MH. Laryngeal carcinoma: high-resolution computed tomography and thick anatomic sections. Radiology 1984;151:689–696. 308. Richards TL, Bowen JD, Alvord EC Jr, et al. Magnetic resonance brain spectroscopy. Basic concepts with emphasis on multiple sclerosis. Int J Neuroradiol 1996;2:123–133. 309. Robinson JD, Crawford SC, Teresi LM, et al. Extracranial lesions of the head and neck: preliminary experience with Gd-DPTA-enhanced MR imaging. Radiology 1989;172:165–170. 310. Rosai J, Carcangiu ML, DeLellis RA. Atlas of tumor pathology: tumors of the thyroid gland. Washington: Armed Forces Institute of Pathology, 1992:19–245. 311. Ross MR, Schomer DF, Chappell P, et al. MR imaging of head and neck tumors: comparison of T1-weighted contrast-enhanced fatsuppressed images with conventional T2-weighted and fast spinecho T2-weighted images. AJR Am J Roentgenol 1994;163:173–178. 312. Rossiter JL, Hendrix RA, Tom LWC, et al. Intramuscular hemangioma of the head and neck. Otolaryngol Head Neck Surg 1993;108:18–26. 313. Rouvier H. Anatomy of the human lymphatic system. Ann Arbor, MI: Edwards Brothers, 1938:1–82. 314. Russell EJ. The radiologic approach to malignant tumors of the head and neck, with emphasis on computed tomography. Clin Plast Surg 1985;12:343–374. 315. Sacks BA, Pallotta JA, Cole A, et al. Diagnosis of parathyroid adenomas: efficacy of measuring parathormone levels in needle aspirates of cervical masses. AJR Am J Roentgenol 1994;163:1223–1226. 316. Sadler TW. Head and neck. In: Sadler TW, ed. Langman’s medical embryology. Baltimore: Williams and Wilkins, 1990:308–313. 317. Sagel SS, AufderHeide JF, Aronberg DJ, et al. High resolution computed tomography in the staging of carcinoma of the larynx. Laryngoscope 1981;9l:292–300. 318. Sakai F, Gamsu G, Dillon WP, et al. MR imaging of the larynx at 1.5T. J Comput Assist Tomogr 1990;14:60–71. 319. Sakai F, Sone S, Kiyono K, et al. MR evaluation of laryngohypopharyngeal cancer: value of gadopentetate dimeglumine enhancement. AJNR Am J Neuroradiol 1993;14:1059–1069. 320. Sakata K, Hareyama M, Tamakawa M, et al. Prognostic factors of nasopharynx tumors investigated by MR imaging and the value of MR imaging in the newly published TNM staging. Int J Radiat Oncol Biol Phys 1999;43:273–278. 321. Salazar JE, Duke RA, Ellis JV. Second branchial cleft cyst: unusual location and a new CT diagnostic sign. AJR Am J Roentgenol 1985;l45:965–966. 322. Saleh EM, Mancuso AA, Alhussaini AA. Computed tomography of primary subglottic cancer: clinical importance of typical spread pattern. Head Neck 1992;14:125–132. 323. Saleh EM, Mancuso AA, Stringer SP. CT of submucosal and occult laryngeal masses. J Comput Assist Tomogr 1992;16:87–93. 324. Saleh EM, Mancuso AA, Stringer SP. Relative roles of computed tomography and endoscopy for determining the inferior extent of pyriform sinus carcinoma: correlative histopathologic study. Head Neck 1993;15:44–52. 325. Sasaki CT, Carlson RD. Malignant neoplasms of the larynx. In: Cummings CW, Frederickson JM, Harker LA, et al., eds. Otolaryngology–head and neck surgery, 2nd ed. St. Louis: CV Mosby, 1993: 1925–1954. 326. Saunders WH. The larynx. Clinical symposia. CIBA Pharmaceutical Co. publication 1964;16:67–99. 327. Saunders WH, Hodgson SE. Bilateral hypoglossal nerve paralysis after irradiation therapy. Ann Otol Rhinol Laryngol 1979;88: 515–517. 328. Schaefer SD, Close LG. Acute management of laryngeal trauma: update. Ann Otol Rhinol Laryngol 1989;98:98–104. 329. Schild JA, Valvassori GE, Mafee MF, et al. Laryngeal malignancies and computerized tomography: a correlation of tomographic and histopathologic findings. Ann Otol Rhinol Laryngol 1982;91:571–575. 330. Scott M, Forsted DH, Rominger CJ, et al. Computed tomographic evaluation of laryngeal neoplasms. Radiology 1981;140:141–144. 331. Shaha AR, LaRosa CA, Jaffe BM. Parathyroid localization prior to primary exploration. Am J Surg 1993;166:289–293.

221

332. Sham JST, Cheung YK, Choy D, et al. Computed tomography evaluation of neck node metastases from nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 1993;26:787–792. 333. Sham JST, Cheung YK, Choy D, et al. Nasopharyngeal carcinoma: CT evaluation of tumor patterns of spread. AJNR Am J Neuroradiol 1991;12:265–270. 334. Shellock FG, Crues JV. High-field-strength MR imaging and metallic biomedical implants: an ex vivo evaluation of deflection forces. AJR Am J Roentgenol 1988;151:389–392. 335. Shpizner BA, Holliday RA. Levator scapulae muscle asymmetry presenting as a palpable neck mass: CT evaluation. AJNR Am J Neuroradiol 1993;14:461–464. 336. Sichel J-Y, Chisin R. Tortuous internal carotid artery: a rare cause of oropharyngeal bulging diagnosed by magnetic resonance angiography. Ann Otol Rhinol Laryngol 1993;102:964–966. 337. Siegel MJ, Glazer HS, St Amour TE, et al. Lymphangiomas in children: MR imaging. Radiology 1989;170:467–470. 338. Sigal R. Infrahyoid neck. Radiol Clin North Am 1998;36:781–799. 339. Sigal R. Oral cavity, oropharynx, and salivary glands. Neuroimaging Clin N Am 1996;6:379–400. 340. Sigal R, Chancelier M-D, Luboinski B, et al. Synovial sarcomas of the head and neck: CT and MR findings. AJNR Am J Neuroradiol 1992;13:1459–1462. 341. Sigal R, Monnet O, de Baere T, et al. Adenoid cystic carcinoma of the head and neck: evaluation with MR imaging and clinicalpathologic correlation in 27 patients. Radiology 1992;184:95–101. 342. Silver AJ, Mawad ME, Hilal SK, et al. Computed tomography of the carotid space and related cervical spaces. Part II: neurogenic tumors. Radiology 1984;l50:729–735. 343. Silver AJ, Mawad ME, Hilal SK, et al. Computed tomography of the carotid space and related cervical spaces. Part I: Anatomy. Radiology 1984;150:723–728. 344. Silver AJ, Mawad ME, Hilal SK, et al. Computed tomography of the nasopharynx and related spaces. Part II: pathology. Radiology 1983;147:733–738. 345. Silverman PM. CT staging of laryngeal and hypopharyngeal carcinoma. In: Glazer GM, ed. Staging of neoplasms. New York: Churchill Livingstone, 1986:55–77. 346. Silverman PM, Bossen EH, Fisher SR, et al. Carcinoma of the larynx and hypopharynx: computed tomographic-histopathologic correlations. Radiology 1984;l5l:697–702. 347. Silverman PM, Johnson GA, Korobkin M. High-resolution sagittal and coronal reformatted CT images of the larynx. AJR Am J Roentgenol l983;140:8l9–822. 348. Silverman PM, Korobkin M. High-resolution computed tomography of the normal larynx. AJR Am J Roentgenol 1983;l40:875–879. 349. Silverman PM, Korobkin M, Moore AV. Computed tomography of cystic neck masses. J Comput Assist Tomogr 1983;7:498–502. 350. Slotnick D, Som PM, Giebfried J, et al. Thyroglossal duct cysts that mimic laryngeal masses. Laryngoscope 1987;97:742–745. 351. Slovis TL, Haller JO, Berdon WE, et al. Current problems in diagnostic radiology: noninvasive visualization of the pediatric airway. Mosby RD, ed. Chicago: Year Book Medical Publishers, 1979: 1–67. 352. Smith AS. Myth of the mesoderm. AJNR Am J Neuroradiol 1989;10:449. 353. Smoker WRK. Normal anatomy of the neck. In: Som PM, Curtin HD, eds. Head and neck imaging, 3rd ed. St. Louis: Mosby Year Book, 1996:711–771. 354. Smoker WRK, Harnsberger HR. Differential diagnosis of head and neck lesions based on their space of origin. 2. The infrahyoid portion of the neck. AJR Am J Roentgenol 1991;157:155–159. 355. Som PM. Cystic lesions of the neck. Postgrad Radiol 1987;7:209–236. 356. Som PM. Detection of metastasis in cervical lymph nodes: CT and MR criteria and differential diagnosis. AJR Am J Roentgenol 1992;158:961–969. 357. Som PM. Lymph nodes of the neck. Radiology 1987;165:593– 600. 358. Som PM. The present controversy over the imaging method of choice for evaluating the soft tissues of the neck. AJNR Am J Neuroradiol 1997;18:1869–1872. 359. Som PM, Biller HF. Computed tomography of the neck in the postoperative patient: radical neck dissection and the myocutaneous flap. Radiology 1983;148:157–160.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 222

222

Chapter 4

360. Som PM, Biller HF. High-grade malignancies of the parotid gland: identification with MR imaging. Radiology 1989;173:823– 826. 361. Som PM, Biller HF, Lawson W, et al. Parapharyngeal space masses: an updated protocol based upon 104 cases. Radiology 1984;153:149–156. 362. Som PM, Brandwein M. Salivary glands. In: Som PM, Curtin HD, eds. Head and neck imaging, 3rd ed. St. Louis: CV Mosby, 1996:823–914. 363. Som PM, Brandwein M, Lidov M, et al. The varied presentations of papillary thyroid carcinoma cervical nodal disease: CT and MR findings. AJNR Am J Neuroradiol 1994;15:1123–1128. 364. Som PM, Braun IF, Shapiro MD, et al. Tumors of the parapharyngeal space and upper neck: MR imaging characteristics. Radiology l987;l64:823–829. 365. Som PM, Curtin HD, Mancuso AA. An imaging-based classification for the cervical nodes designed as an adjunct to recent clinically based nodal classifications. Arch Otolaryngol Head Neck Surg 1999;125:388–396. 366. Som PM, Sacher M, Lanzieri CF, et al. Two benign CT presentations of thyroid-related papillary adenocarcinoma. J Comput Assist Tomogr 1985;9:162–166. 367. Som PM, Sacher M, Lanzieri CF, et al. Parenchymal cysts of the lower neck. Radiology l985;157:399–406. 368. Som PM, Sacher M, Stollman AL, et al. Common tumors of the parapharyngeal space: refined imaging diagnosis. Radiology 1988;169:81–85. 369. Som PM, Scherl MP, Rao VM, et al. Rare presentations of ordinary lipomas of the head and neck: a review. AJNR Am J Neuroradiol 1986;7:657–664. 370. Som PM, Urken ML, Biller H, et al. Imaging the postoperative neck. Radiology 1993;187:593–603. 371. Soucek M, Vock P, Daepp M, et al. [Spiral CT: a new technique for volumetric scans. II. Potential clinical applications.] Röntgenpraxis 1990;43:365–375. German. 372. Spreer J, Krahe T, Jung G, et al. Spiral versus conventional CT in routine examinations of the neck. J Comput Assist Tomogr 1995;19:905–910. 373. Spritzer CE, Gefter WB, Hamilton R, et al. Abnormal parathyroid glands: high-resolution MR imaging. Radiology l987;l62:487–49l. 374. Stanley RE. Parapharyngeal space tumours. Ann Acad Med Singapore 1991; 20:589–596. 375. Stark DD, Gooding GAW, Clark OH. Noninvasive parathyroid imaging. Semin Ultrasound CT MR l985;6:3l0–320. 376. Stark DD, Moss AA, Gamsu G, et al. Magnetic resonance imaging of the neck. Part I. Normal anatomy. Radiology 1984;150:447– 454. 377. Stark DD, Moss AA, Gamsu G, et al. Magnetic resonance imaging of the neck. Part II. Pathologic findings. Radiology 1984;150: 455– 461. 378. Steinkamp HJ, Hosten N, Richter C, et al. Enlarged cervical lymph nodes at helical CT. Radiology 1994;191:795–798. 379. Stern WBR, Silver CE, Zeifer BA, et al. Computed tomography of the clinically negative neck. Head Neck 1990;12:109–113. 380. Stevens MH, Harnsberger HR, Mancuso AA, et al. Computed tomography of cervical lymph nodes: staging and management of head and neck cancer. Arch Otolaryngol 1985;111:735–739. 381. Stewart MG, Hayman LA, Taber KH, et al. Clinical pathology of the neck: spatial compartments. Int J Neuroradiol 1998;4:152– 158. 382. Suh J-S, Abenoza P, Galloway HR, et al. Peripheral (extracranial) nerve tumors: correlation of MR imaging and histologic findings. Radiology 1992;183:341–346. 383. Sulfaro S, Barzan L, Querin F, et al. T staging of the laryngohypopharyngeal carcinoma. A 7-year multidisciplinary experience. Arch Otolaryngol Head Neck Surg 1989;115:613–620. 384. Suojanen JN, Mukherji SK, Dupuy DE, et al. Spiral CT in evaluation of head and neck lesions: work in progress. Radiology 1992;183:281–283. 385. Suojanen JN, Mukherji SK, Wippold FJ. Spiral CT of the larynx. AJNR Am J Neuroradiol 1994;15:1579–1582. 386. Swartz JD, Rothman MI, Marlowe FI, et al. MR imaging of parotid mass lesions: attempts at histopathologic differentiation. J Comput Assist Tomogr 1989;13:789–796.

387. Swartz JD, Yussen PS, Popky GL. Imaging the soft tissues of the neck: non-nodal acquired disease. Crit Rev Diagn Imaging 1991;31:471–547. 388. Taber KH, Hayman LA, Diaz-Marchan PJ, et al. Spatial compartments of the neck, part II: sagittal sections. Int J Neuroradiol 1998;4:313–319. 389. Takashima S, Fukuda H, Tomiyama N, et al. Hashimoto thyroiditis: correlation of MR imaging signal intensity with histopathologic findings and thyroid function test results. Radiology 1995;197:213–219. 390. Takashima S, Ikezoe J, Harada K, et al. Tongue cancer: correlation of MR imaging and sonography with pathology. AJNR Am J Neuroradiol 1989;10:419–424. 391. Takashima S, Morimoto S, Ikezoe J, et al. CT evaluation of anaplastic thyroid carcinoma. AJNR Am J Neuroradiol 1990;11: 361–367. 392. Takashima S, Noguchi Y, Takeuchi N, et al. Head and neck carcinoma: detection of extraorgan spread with MR imaging and CT. Eur J Radiol 1992;14:228–234. 393. Takashima S, Nomura N, Noguchi Y, et al. Primary thyroid lymphoma: evaluation with US, CT, and MRI. J Comput Assist Tomogr 1995;19:282–288. 394. Tamler BJ, Kligerman MM. Diagnostic imaging aids to head and neck radiation oncology. Neuroimaging Clin N Am 1996;6:515– 530. 395. Tampieri D, Melanson D, Ethier R. MR imaging of epidermoid cysts. AJNR Am J Neuroradiol 1989;10:351–356. 396. Tart RP, Mukherji SK, Lee WR, et al. Value of laryngeal cartilage sclerosis as a predictor of outcome in patients with stage T3 glottic cancer treated with radiation therapy. Radiology 1994;192: 567–570. 397. Tartaglino LM, Rao VM, Markiewicz DA. Imaging of radiation changes in the head and neck. Semin Roentgenol 1994;29: 81–91. 398. Tavill MA, Poje CP, Wetmore RF, et al. Plunging ranulas in children. Ann Otol Rhinol Laryngol 1995;104:405–408. 399. Teitelbaum GP, Lin MCW, Watanabe AT, et al. Ferromagnetism and MR imaging: safety of carotid vascular clamps. AJNR Am J Neuroradiol 1990;11:267–272. 400. Teresi LM, Lufkin RB, Vinuela F, et al. MR imaging of the nasopharynx and floor of the middle cranial fossa. Part I. Normal anatomy. Radiology 1987;164:811–816. 401. Teresi LM, Lufkin RB, Vinuela F, et al. MR imaging of the nasopharynx and floor of the middle cranial fossa. Part II. Malignant tumors. Radiology 1987;164:817–821. 402. Teresi LM, Lufkin RB, Wortham DG, et al. Parotid masses: MR imaging. Radiology 1987;163:405–409. 403. Terris DJ, Arnstein DP, Nguyen HH. Contemporary evaluation of unilateral vocal cord paralysis. Otolaryngol Head Neck Surg 1992;107:84–90. 404. Tien RD, Hesselink JR, Chu PK, et al. Improved detection and delineation of head and neck lesions with fat suppression spinecho MR imaging. AJNR Am J Neuroradiol 1991;12:19–24. 405. Tomiak MM, Foley WD, Jacobson DR. Variable-mode helical CT: imaging protocols. AJR Am J Roentgenol 1995;164:1525–1531. 406. Towler CR, Young SW. Magnetic resonance imaging of the larynx. Magn Reson Q 1989;5:228–241. 407. Tryhus MR, Smoker WRK, Harnsberger HR. The normal and diseased masticator space. Semin Ultrasound CT MR 1990;11:476– 485. 408. Tucker GF Jr. The anatomy laryngeal cancer. Can J Otolaryngol l974;3:417–431. 409. Unger JM. The oral cavity and tongue: magnetic resonance imaging. Radiology 1985;155:151–153. 410. van den Brekel MWM, Castelijns JA. Radiologic evaluation of neck metastases: the otolaryngologist’s perspective. Semin Ultrasound CT MR 1999;20:162–174. 411. van den Brekel MWM, Castelijns JA, Snow GB. Imaging of cervical lymphadenopathy. Neuroimaging Clin N Am 1996;6:417– 434. 412. van den Brekel MWM, Castelijns JA, Snow GB. The role of modern imaging studies in staging and therapy of head and neck neoplasms. Semin Oncol 1994;21:340–348. 413. van den Brekel MWM, Castelijns JA, Stel HV, et al. Detection and characterization of metastatic cervical adenopathy by MR imag-

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 223

Neck ing: comparison of different MR techniques. J Comput Assist Tomogr 1990;14:581–589. 414. van den Brekel MWM, Stel HV, Castelijns JA, et al. Cervical lymph node metastasis: assessment of radiologic criteria. Radiology 1990;177:379–384. 415. van Dijke CF, van Waes PFGM. Head and neck tumors, MRI versus CT: a technology assessment pilot study. Eur J Radiol 1992; 14:235–239. 416. van Vierzen PBJ, Joosten FBM, Manni JJ. Sonographic, MR and CT findings in a large laryngocele: a case report. Eur J Radiol 1994;18:45–47. 417. Varma DGK, Moulopoulos A, Sara AS, et al. MR imaging of extracranial nerve sheath tumors. J Comput Assist Tomogr 1992;16: 448–453. 418. Vogl T, Dresel S, Bilaniuk LT, et al. Tumors of the nasopharynx and adjacent areas: MR imaging with Gd-DTPA. AJNR Am J Neuroradiol 1990;11:187–194. 419. Vogl T, Dresel S, Juergens M, et al. MR imaging with Gd-DTPA in lesions of the head and neck. J Otolaryngol 1993;22:220–230. 420. Vogl T, Wilimzig C, Grevers G, et al. 3D MR imaging with GdDTPA in head and neck lesions. Eur Radiol 1991;1:151–157. 421. Vogl TJ, Balzer JO. Base of skull, nasopharynx, and parapharyngeal space. Neuroimaging Clin N Am 1996;6:357–378. 422. Vogl TJ, Mack MG, Juergens M, et al. MR diagnosis of head and neck tumors: comparison of contrast enhancement with tripledose gadodiamide and standard-dose gadopentetate dimeglumine in the same patients. AJR Am J Roentgenol 1994;163:425–432. 423. Vogl TJ, Steger W, Balzer JO, et al. MRI of the hypopharynx, larynx and neck. Eur Radiol 1992;2:391–408. 424. Vogl TJ, Steger W, Ihrler S, et al. Cystic masses in the floor of the mouth: value of MR imaging in planning surgery. AJR Am J Roentgenol 1993;161:183–186. 425. Vogler RC, Wippold FJ 2nd, Pilgram TK. Age-specific size of the normal adenoid pad on magnetic resonance imaging. Clin Otolaryngol 2000;25:392–395. 426. Vokes EE, Weichselbaum RR, Lippman SM, et al. Head and neck cancer. N Engl J Med 1993;328:184–194. 427. Wang J-C, Takashima S, Takayama F, et al. Tracheal invasion by thyroid carcinoma: prediction using MR imaging. AJR Am J Roentgenol 2001;177:929–936. 428. Ward MP, Glazer HS, Heiken JP, et al. Traumatic perforation of the pyriform sinus: CT demonstration. J Comput Assist Tomogr l985;9:982–984. 429. Watarai J, Seino Y, Kobayashi M, et al. CT of retropharyngeal lymph node metastasis from maxillary carcinoma. Acta Radiol 1993;34:492–495. 430. Weinberger MS, Robbins K. Diagnostic localization studies for primary hyperparathyroidism. A suggested algorithm. Arch Otolaryngol Head Neck Surg 1994;120:1187–1189. 431. Weinstein GS, Laccourreye O, Brasnu D, et al. The role of computed tomography and magnetic resonance imaging in planning for conservation laryngeal surgery. Neuroimaging Clin N Am 1996;6:497–504. 432. Weissman JL, Curtin HD. Benign erosion of laryngeal cartilage: report of two cases. AJNR Am J Neuroradiol 1990;11:1215–1216. 433. Weissman JL, Curtin HD, Johnson JT. Thyroid gland after total laryngectomy: CT appearance. Radiology 1998;207:405–409. 434. Welsh LW, Welsh JJ, Gregor FA. Radiographic analysis of deep cervical abscesses. Ann Otol Rhinol Laryngol 1992;101:854–860. 435. Wenig BM. Neoplasms of the salivary glands. In: Wenig BM. Atlas of head and neck pathology. Philadelphia: WB Saunders, 1993:283–334. 436. Wenig BM. Non-neoplastic lesions of the oral cavity, nasopharynx, tonsils, and neck. In: Atlas of head and neck pathology. Philadelphia: WB Saunders, 1993:105–141. 437. Wester DJ, Whiteman MLH, Singer S, et al. Imaging of the postoperative neck with emphasis on surgical flaps and their complications. AJR Am J Roentgenol 1995;164:989–993. 438. Williams DW 3rd. An imager’s guide to normal neck anatomy. Semin Ultrasound CT MR 1997;18:157–181. 439. Willinsky RA, Kassel EE, Cooper PW, et al. Computed tomography of lingual thyroid. J Comput Assist Tomogr 1987;11:182–183.

223

440. Wilner HI, Lazo A, Metes JJ, et al. Embolization in cataclysmal hemorrhage caused by squamous cell carcinomas of the head and neck. Radiology 1987;163:759–762. 441. Wippold FJ 2nd. Postoperative pharynx. In: Gore RM, Levine MS, eds. Textbook of gastrointestinal radiology, 2nd ed. Philadelphia: WB Saunders, 2000;257–270. 442. Wippold FJ 2nd, Glazer HS, Siegel MJ. Imaging of the larynx. In: Cummings CW, Fredrickson JM, Harker LA, et al., eds. Otolaryngology–head and neck surgery, 2nd ed. St. Louis: CV Mosby, 1993;1785–1805. 443. Wippold FJ 2nd, Heiken J. Imaging of the hypopharynx. In: Cummings CW, Frederickson JM, Harker LA, et al., eds. Otolaryngology–head and neck surgery, 2nd ed. St. Louis: CV Mosby, 1993; 1806–1830. 444. Wippold FJ 2nd, Smirniotopoulos JG, Moran CJ, et al. Chondrosarcoma of the larynx: CT features. AJNR Am J Neuroradiol 1993;14:453–459. 445. Wolff SD, Balaban RS. Magnetization transfer contrast (MTC) and tissue water proton relaxation in vivo. Magn Reson Med 1989;10:135–144. 446. Woodruff WW, Kennedy TL. Non-nodal neck masses. Semin Ultrasound CT MR 1997;18:182–204. 447. Wortham DG, Hoover LA, Lufkin RB, et al. Magnetic resonance imaging of the larynx: a correlation with histologic sections. Otolaryngol Head Neck Surg 1986;94:123–133. 448. Yeager VL, Archer CR. Anatomical routes for cancer invasion of laryngeal cartilages. Laryngoscope 1982;92:449–452. 449. Yeager VL, Lawson C, Archer CR. Ossification of the laryngeal cartilages as it relates to computed tomography. Invest Radiol l982;1:11–19. 450. Yousem DM. Dashed hopes for MR imaging of the head and neck: the power of the needle. Radiology 1992;184:25–26. 451. Yousem DM. Parathyroid and thyroid imaging. Neuroimaging Clin N Am 1996;6:435–459. 452. Yousem DM. Suprahyoid spaces of the head and neck. Semin Roentgenol 2000;35:63–71. 453. Yousem DM, Chalian AA. Oral cavity and pharynx. Radiol Clin North Am 1998;36:967–981. 454. Yousem DM, Hatabu H, Hurst RW, et al. Carotid artery invasion by head and neck masses: prediction with MR imaging. Radiology 1995;195:715–720. 455. Yousem DM, Hurst RW. MR of cervical lymph nodes: comparison of fast spin-echo and conventional spin-echo T2W scans. Clin Radiol 1994;49: 670–675. 456. Yousem DM, Montone KT, Sheppard LM, et al. Head and neck neoplasms: magnetization transfer analysis. Radiology 1994;192: 703–707. 457. Yousem DM, Sack MJ, Scanlan KA. Biopsy of parapharyngeal space lesions. Radiology 1994;193:619–622. 458. Yousem DM, Scheff AM. Thyroid and parathyroid. In: Som PM, Curtin HD, eds. Head and neck imaging, 3rd ed. St. Louis: Mosby, 1996:952–975. 459. Yousem DM, Som PM, Hackney DB, et al. Central nodal necrosis and extracapsular neoplastic spread in cervical lymph nodes: MR imaging versus CT. Radiology 1992;182:753–759. 460. Yuhan JH, Hayman LA, Taber KH, et al. The larynx: normal anatomy and surgical considerations. Int J Neuroradiol 1997;3: 515–520. 461. Zadvinskis DP, Benson MT, Kerr HH, et al. Congenital malformations of the cervicothoracic lymphatic system: embryology and pathogenesis. Radiographics 1992;12:1175–1189. 462. Zeiberg AS, Silverman PM, Sessions RB, et al. Helical (spiral) CT of the upper airway with three-dimensional imaging: technique and clinical assessment. AJR Am J Roentgenol 1996;166:293–299. 463. Zoarski GH, Lufkin RB, Bradley WG Jr, et al. Multicenter trial of gadoteridol, a nonionic gadolinium chelate, in patients with suspected head and neck pathology. AJNR Am J Neuroradiol 1993;14:955–961. 464. Zoarski GH, Mackey JK, Anzai Y, et al. Head and neck: initial clinical experience with fast spin-echo MR imaging. Radiology 1993;188:323–327.

5063_Lee_Ch04pp0145-0224 10/13/05 12:07 PM Page 224

5063_Lee_Ch05pp0225-0310 10/14/05 2:08 PM Page 225

Thorax: Techniques and Normal Anatomy Fernando R. Gutierrez

Santiago Rossi

Although chest radiography remains the primary radiological method for investigating diseases of the thorax, the many advantages provided by computed tomography (CT) and magnetic resonance imaging (MRI) are well established. These include multiplanar depiction of anatomy, superior contrast sensitivity, and ability to view the entire dynamic range of lung, fat, soft tissue, water, and bone densities inherent in the thorax. As with chest radiography, soft copy viewing coupled with picture archive and communication systems (PACS) allows image manipulation and remote viewing not only by radiologists but also by clinicians, thus enhancing the medical care of these patients (226). Abnormalities found or suspected on standard radiographs may be characterized to better advantage by the ability of CT or MR to display overlapping structures in a cross-sectional format (38). The detection of relatively small lesions is enhanced by the manner in which CT separates structures that would be superimposed on standard radiographic projections (249). In addition, this information can be displayed with three-dimensional (3D) volume rendering to provide a unique perspective on thoracic anatomy and disease (71,152). The strength of CT over conventional radiography resides in contrast resolution, which is the ability to discriminate small differences in tissue attenuation. Gas, fat, water, soft tissue, osseous and metallic densities are more easily and clearly discriminated (215). Small adjacent structures of differing densities are readily distinguishable. Relatively low concentrations of intravenously administered contrast material are capable of opacifying vascular structures, depicting anatomic information that can be extremely crucial for treatment planning in certain cases.

5

Sanjeev Bhalla

Spiral CT, introduced in the early 1990s led to an increase in the utilization of CT, a revolution that continued with the advent of multidetector row CT (MDCT) scanners more recently (183,203). The speed of helical technology has drastically reduced the time required to image an area of anatomic interest. The net result has been an increased number of applications in seriously ill patients and in those in which a long region of anatomy needs to be scanned in a single breath-hold (199). Also, helical technology permits multiple phases of scanning to be obtained after one contrast injection (Fig. 5-1). As a consequence, the depiction of thoracic anatomy achieved with CT and magnetic resonance imaging (MRI) has transformed the diagnostic investigation of potential chest pathology. No longer is a linear, traditional approach always the best use of medical resources, i.e., a single chief complaint followed by the history and physical examination performed in the hospital, followed by a deliberate, written consideration of which tests might be most directive. As part of this change in medicine, it is not uncommon to have the chest CT or MR examination requested as an initial study of an in-hospital evaluation of a possible chest disease (e.g., suspected aortic dissection) or as a follow-up of an abnormality seen on an admitting chest radiograph. A common scenario is to have a chest CT requested following an unsuspected abnormal, but nonspecific chest radiographic finding during an admission or evaluation for an unrelated problem (Fig. 5-2). As a result of their inherent advantages and speed, crosssectional imaging has become increasingly utilized. As a consequence, it has become necessary for imaging specialists to become active participants in the diagnostic process; involved in assessing the appropriateness of particular tests

225

5063_Lee_Ch05pp0225-0310 10/14/05 2:08 PM Page 226

226

Chapter 5

A

B

Figure 5-1 A: Non-contrast CT in a 68-year-old man with chest pain. B: Standard arterial phase scan was obtained using the aortic dissection protocol (fixed delay 35 seconds) on a 4-row multidetector computed tomography (MDCT). A Stanford type A aortic dissection was discovered, as well as hemopericardium. The false lumen (arrow) does not enhance on this phase. C: Because of the speed of MDCT, a venous phase scan could be obtained (delay 90 seconds). The added value of this phase is that it enables flow in the false lumen to be seen. This additional phase was possible because the helical technology allowed the entire aorta to be scanned in 25 seconds.

C

and triaging the order of patients. A thorough knowledge of the potential benefits and risks, as well as the specific types of information provided by these cross-sectional chest imaging studies is a prerequisite (53,127,191). Although in general the benefits of CT tend to outweigh the risks, one particular area of concern has been the radiation dose and its carcinogenic effects, particularly in pediatric applications (248). The specificity of both CT and MR is related, in part, to their cross-sectional format, allowing visualization and differentiation of structures superimposed on plain chest radiography. The enhanced contrast sensitivity of crosssectional studies results in a more than 10-fold improvement in tissue differentiation over standard radiographs. Cross-sectional imaging is especially advantageous in those chest regions that are particularly challenging for plain chest radiography: the mediastinum, perimediastinal lung, the peridiaphragmatic areas and the subpleural regions. Such tissue characterization by CT and MRI may be crucial in differentiating between a benign and a malignant

process (Fig. 5-3). Unfortunately, in many instances, and with the exemption of cardiovascular conditions, MRI has failed to fulfill its initial promise as a more specific means than CT in discriminating types of pathologic processes in the thorax. Furthermore, diagnostic quality CT images can be obtained more rapidly and easily in virtually all patients, even those who are dependent upon various intensive care monitoring and life support devices. Even in the cardiovascular arena, newer developments of MDCT have demonstrated the usefulness of CT in obtaining good quality studies in a short period of time, creating a robust alternative for this type of application. The ability of CT to distinguish various tissues and organs found in the chest, characterizing them as normal or abnormal, has produced major benefit in directing or obviating further studies and invasive procedures (54). Based on the chest radiograph alone, there may be difficulty in distinguishing mediastinal, lung and chest wall lesions. Such tissue characterization may be crucial in differentiating between a benign process and malignant disease.

5063_Lee_Ch05pp0225-0310 10/14/05 2:08 PM Page 227

Thorax: Techniques and Normal Anatomy

A

B

C

D

227

Figure 5-2 A, B: A 56-year-old patient complaining of dyspnea. Standard chest radiograph demonstrates subcarinal soft tissue mass (arrows). Axial (C) and coronal (D) images of contrastenhanced CT reveals cystic nature of the mass (M). Incidentally noted are extensive bilateral pulmonary thromboemboli (arrows), explaining patient’s symptoms.

INDICATIONS There is now wide acceptance of chest CT in characterizing the nature and extent of most thoracic tumors and lung diseases, while MRI has established a role in the evaluation of certain cardiovascular lesions, superior sulcus tumors, chest wall abnormalities and thoracic masses abutting the spine, with potential for involvement of the spinal cord. The cardiovascular applications of CT and MRI are addressed in a separate chapter of this book. One of the most common indications for CT is in the further evaluation of a pulmonary nodule (Fig. 5-4) or suspected neoplasm, both at its initial presentation and for follow-up after therapy (surgical, radiation or chemotherapy). Evaluation of bronchogenic carcinoma comprises the majority of these cases. Other primary chest neoplasms, including

lymphoma and sarcoma, as well as potential metastatic disease, are also frequently evaluated with CT. Magnetic resonance imaging has a number of applications in the thorax and these have been gradually increasing with promising new developments (143,274). MR recognition of flowing blood, high contrast resolution of soft tissue structures, and the absence of ionizing radiation make MR an attractive modality for examining the chest. However, several technical obstacles limit MR from becoming the routine first-line tomographic technique of choice in the thorax. Although newer pulse sequences have made inroads in the study of lung parenchyma (22,47), particularly in the assessment of perfusion and ventilation (46,107,161,184). MR imaging currently has limited clinical value in this area. The poor signal potential of the aerated lung parenchyma is perhaps the most significant obstacle to successful MR imaging of this

5063_Lee_Ch05pp0225-0310 10/14/05 2:08 PM Page 228

228

Chapter 5

B

A Figure 5-3 A: Black blood HASTE (half-Fourier acquisition single-shot turbo spin-echo) magnetic resonance images that clearly depict different signal intensities of the different structures of the thorax. Abundant chest wall and mediastinal fat surrounds vascular structures and the tracheobronchial tree. B: The heart is also seen surrounded by large amount of adipose tissue. Note lack of signal intensity inside heart chambers. Myocardial tissue is seen as intermediate signal intensity. The study was obtained because of an enlarged mediastinal silhouette seen on a routine chest radiograph.

organ (22). As the lungs are mostly air, with the density of lung parenchyma averaging 0.23 g/mL, less than any other organ in the body (22), there is an inherent low proton density and relatively short transverse relaxation time (T2). In addition, multiple air-tissue interfaces, the defining feature of lung parenchyma, tend to induce local gradients, which

affect magnetic field homogeneity, resulting in further signal loss and magnetic susceptibility effects (22). Respiratory motion compounds the problem of generating high quality anatomic images of the chest, although newly developed gating techniques have proven effective. Recent work with hyperpolarized gases, namely helium, shows early promise

A

B Figure 5-4 A: Postcontrast enhanced computed tomography (CT) in a 68-year-old man with a history of renal cell carcinoma. CT was performed because of a questionable nodule on chest radiography. Lung window shows a smooth, well-circumscribed nodule in the right lower lobe. B: Softtissue window clearly shows central fat attenuation. The combination of smooth borders and fat attenuation are diagnostic of a pulmonary hamartoma.

5063_Lee_Ch05pp0225-0310 10/14/05 2:08 PM Page 229

Thorax: Techniques and Normal Anatomy

229

A

B Figure 5-5 A: Postcontrast T1-weighted magnetic resonance image through the lung base at the level of the fossa ovalis. Pulmonary arteries are easily seen, but underlying parenchyma is difficult to assess. B: CT without intravenous contrast at the same level as A, using the same slice thickness (5 mm). Few septal lines and areas of patchy ground glass are much easier to appreciate. In this case, findings were related to asbestosis.

in increasing the temporal and spatial resolution of MR imaging of airspaces and may result in a greater role in the MR evaluation of the lungs (5). Some recent studies have used dynamic contrast-enhanced MR imaging with use of kinetic and morphologic parameters in the evaluation of pulmonary nodules and pulmonary ventilation with some success (129,235); however, CT remains the workhorse in these important diagnostic tasks. Finally, the very low signal of calcium is a disadvantage in identifying this important component of many intrathoracic lesions. The pathognomic findings of bulky calcifications in hamartomas or of densely calcified lymph nodes or parenchymal nodules of granulomatous disease, so easily recognized on CT, may be missed on the MR examination. These factors, viewed alongside CT’s greater speed, lesser cost and more general availability, make MR a far less attractive technique for the routine imaging of lung parenchyma (Fig. 5-5). From a practical clinical perspective, and taking advantage of its inherent high contrast resolution and sensitivity to flowing blood, magnetic resonance imaging is often used to answer specific questions in the thorax raised by other imaging modalities. It may be utilized to assess extent of tumor invasion into the mediastinum and great vessels, to study possible vascular lesions within the chest and in the assessment of pulmonary or mediastinal lesions in which fibrosis is a consideration. Specific clinical indications for which magnetic resonance imaging may be valuable include suspected aortic disease (dissection, aneurysms, coarctation), superior sulcus tumors, brachial plexus masses or trauma, paramediastinal and mediastinal masses (Fig. 5-6), juxtadiaphragmatic masses, chest wall lesions and central pulmonary arterial disorders including chronic thromboemboli (46,85).

COMPUTED TOMOGRAPHY AND MAGNETIC RESONANCE TECHNIQUES Preliminaries: Preparing to Scan To increase patient throughput and decrease patient anxiety, the actual time spent in the scanning suite must be reduced to a minimum (33). This is best accomplished by completing as many tasks outside of the CT or MR scanning suite as is reasonably possible (209). These tasks include patient interview, obtaining informed consent and planning an imaging strategy. The radiologic technologist is made aware of the specific imaging protocols that are needed for the particular case. One method of facilitating the technologist’s task is to standardize examination techniques whenever possible. When possible, on-line monitoring of the examination permits quick modification of the technique and helps avoid unnecessary repetition. Ready availability of the radiologist is of prime importance to running a smooth operation. As it is essential to understand the context of the current study, when the clinical problem to be addressed is not well defined by review of the written requisition, through the patient interview or review of the medical chart, a telephone call to the referring physician is often warranted. It is often necessary to tailor the examination to best address the current clinical question. A preliminary discussion with the patient is often advantageous in reducing anxiety, determining the need for and type of contrast material to be used, and in the case of MR, for discerning any contraindications to imaging. Generally, our patient workload is divided into two broad groups: inpatients and outpatients. Unless previously specified, all patients are released from our department

5063_Lee_Ch05pp0225-0310 10/14/05 2:08 PM Page 230

230

Chapter 5

A

B

C

D Figure 5-6 A: Chest radiograph in a 48-year-old man with a long history of smoking and a chronic cough shows a central mass that turned out to be a small cell carcinoma of the lung. B: Magnetic resonance imaging was performed to evaluate the relationship of the mass to the mediastinal structures. Coronal half-Fourier acquisition single-shot turbo-spin-echo (HASTE) image shows the mass (M) invading the mediastinum and the right pulmonary artery (PA). C: Sagittal HASTE image shows the mass within the right pulmonary artery (arrow). D: Axial postcontrast fat-suppressed 3D gradient recalled echo (GRE) sequence shows the mass as a filling defect within the right pulmonary artery (M).

prior to CT image review by the radiologist. When the potential need for additional images beyond the standard protocol is suggested by the clinical history, or when coming for the exam from a great distance, patients are asked to remain in the imaging area until a preliminary review of the images is completed. Another exception to immediate release after scanning is the intensive care unit patient, in whom studies are reviewed while the patient remains on

the scanning table. In these critically ill patients, monitoring devices should be positioned in view of the control area to allow for continuous observation. Where possible, the ability of the radiologist to review images from a console in the reading area outside of the CT control room increases efficiency. Close attention to technique also is important for maximum productivity in MR. The principal factor that deter-

5063_Lee_Ch05pp0225-0310 10/14/05 2:08 PM Page 231

Thorax: Techniques and Normal Anatomy

mines MR throughput is the duration of the MR examination, i.e., the number of sequences performed (182). Standard protocols dictating the number and type of sequences have proven efficient, with frequent re-evaluation of the sequences as the technology changes (67,96). The practice of reviewing one sequence before performing the next should be discouraged for all but the most unusual and complex cases. By using established and predetermined protocols, throughput rates are said to increase by 80% to 125% (182). The prospective selection of patients for whom a preplanned examination can be used will also increase throughput (182). This preplanning includes clustering similar exams together to minimize the number of times the room and examination table must be customized for the next patient. Screening prior to the MR examination is needed to ensure patient safety. Each patient should be both provided with a questionnaire listing specific contraindications to MR imaging and be personally questioned regarding prior surgeries, injuries or metallic implants. Specifically, patients with ferromagnetic intracerebral aneurysm clips, pacemakers, implantable cardiac defibrillators, intraocular metal fragments or cochlear implants should not be imaged with MR. Patients who give a history of sheet metal work and eye injury should undergo a pre-MR set of orbital plain xray films (244). Patients with pacemakers should not be imaged under any circumstance, not only because of the possibility of torquing the apparatus, but also because of the possibility of inducing currents in the pacer leads which could produce undesired rhythms or soft-tissue thermal injury. Patients with intracranial aneurysm clips should not be imaged without documentation that the clip is nonferromagnetic. While most orthopedic devices can be imaged, scanning should be discontinued if the patient reports feeling warm or hot in the area of the implant (244). Preliminary brief discussions with the patient are also necessary in order to review the performance of the study, including instructions for breath-hold images and communication with the technologist during the examination. Since the enclosed and somewhat confined space within the magnet bore may elicit claustrophobia and the noise of the machine may be daunting, it is important to find ways to increase patient tolerance and acceptance. Anxiety and panic attacks during magnetic resonance scans are not uncommon (33). Between 10% and 20% of patients cannot complete MR imaging because of claustrophobic reactions (139). Either earphones or earplugs are extremely helpful in protecting patients from machine noise in MR. Music through the former makes the experience more pleasant. For patients that have a history of claustrophobia, a sedative may be helpful prior to the examination. Traditionally, patient monitoring has been more difficult for MR imaging than for CT. The development of MRcompatible monitoring devices, intravenous pumps and even ventilators has facilitated MR imaging of the clinically unstable patient.

231

Helical/Spiral CT The development of helical (spiral) CT technology has had profound impact on the imaging of thoracic diseases (144,250). The slip-ring gantry allows for continuous rotation of the x-ray source and detector system. Coupled with continuous movement of the imaging table, the x-ray beam traces a helical or spiral path about the patient, resulting in the acquisition of uninterrupted volumetric data. The images generated by helical technique must be reformatted by interpolation into the standard axial plane and are comparable to those obtained by conventional CT scanning (271). In the thorax, the major advance afforded by helical technique relates to the suppression of respiratory misregistration artifacts. The entire thorax can be imaged during a single breath-hold, without the slice-to-slice variation in lung volume inherent to conventional dynamic scanning. Other advantages of helical technique over conventional CT include the ability to perform reconstructions at small and retrospectively selectable increments, increased scanning speed, decreased motion artifact, optimized utilization of contrast material, flexible multiplanar two-dimensional (2D) reconstruction and vastly improved 3D reconstruction capabilities (39,220). The ability to manipulate volumetric data sets acquired in helical scanning has led to new applications in thoracic CT angiography and airways imaging.

Computed Tomography Scan Parameters Selecting Technique Patient Positioning CT of the thorax is most commonly performed in the supine position. The arms are elevated above the head to reduce beam-hardening artifact from the shoulders and upper extremities. In the supine position, a gradient of increasing attenuation is seen from the anterior to the posterior portions of the lung, owing to relatively diminished inflation and increased blood flow to the more dependent posterior (dorsal) lung parenchyma (267,288). In the prone position, this physiologic phenomenon and its associated attenuation gradient are reversed. Thus, on occasions when the posterior lungs are of primary interest, such as the evaluation of subtle interstitial lung disease, prone positioning may be useful (Fig. 5-7) (259). Similarly, areas of increased opacity at the posterior lung bases, thought to represent dependent compressive atelectasis, can be identified with confidence by prompt resolution when the patient is imaged in the prone position. Lateral decubitus imaging is occasionally valuable to demonstrate the positional dependency of material within a cavity, or in distinguishing a complex pleural process from a parenchymal abnormality (e.g., empyema from lung abscess). A prominent gastroesophageal junction may be differentiated from a mass by imaging in the left lateral decubitus or prone position.

5063_Lee_Ch05pp0225-0310 2/17/06 9:43 AM Page 232

232

Chapter 5

A

B Figure 5-7 A: Subtle fibrosis in the dependent portion of the right lung (arrow) could easily be confused for atelectasis on this noncontrast CT examination. B: Prone images confirm that this area (arrow) represents fibrosis. Atelectasis would not remain present.

Preliminary Digital Radiograph On CT, the scout image, scanogram or topogram consists of a computed radiograph that is reconstructed from numerous contiguous images with the x-ray tube held in a fixed position. In CT of the thorax, the upper abdomen is almost always included. A lateral topogram may be used when the goal is to limit scanning to the lungs only, as with lung screening. MR examination, however, generally focuses on a targeted area of the thorax with the area of interest in the center of the magnet. The selection of specific CT imaging techniques in the thorax has become increasingly complex. The challenge is compounded by expanding applications and indications for CT imaging in thoracic disease and the greatly varied capabilities of available equipment (268). No single method is optimal for all studies in all patients. The information depicted on thin-section high-resolution computed tomography of the pulmonary parenchyma is vastly different from that obtained by conventional, contiguous or helical scanning of the entire thoracic volume). The advances provided by helical scan technique require the selection of a host of new image acquisition and reconstruction parameters. Optimized contrast administration techniques are essential in the depiction of vascular anatomy and pathology, especially in the case of specialized indications such as the detection of pulmonary embolism or aortic dissection (Fig. 5-8).

It is important to note that for conventional CT imaging, slice thickness has significant impact on partial volume averaging effects. The thinner the slice, the less partial volume effect appears in the image. MDCT has further complicated the selection of collimation. Depending on the number of channels and the type of machine, the technologist can set the collimation of the scanner anywhere from submillimeter to 5 millimeters. Unlike single-detector CT, the slice thickness is less dependent on collimation and slice thickness is less susceptible to slice broadening (78). The only requisite is that the slice thickness must be greater than the selected collimation. Choosing the right collimation becomes a balance between radiation dose (which increases as the collimation is reduced), speed (which decreases as collimation is reduced) and spatial resolution (which increases as collimation is reduced). In routine abdominal studies, where thin sections are rarely needed, the collimation can be set from 3-5 mm, so that radiation dose is minimized and speed is maximized. For routine chest CT, a smaller collimation may be used (1 to 2 mm) so that slices may be displayed as 5 mm and reconstructed as 1 mm thick should a nodule or ground-glass opacity be encountered. To keep radiation dose low, the mAs, kVp may be lowered or an on-line tube modulation program may be used (118). The advantage of MDCT is that the thinner images may be reconstructed without rescanning the patient, as long as the helical data is still present.

Scan Collimation Single-detector helical chest CT examinations are usually performed with 5- to 8-mm collimation, a pitch of 1, and corresponding 5- to 8-mm data reconstruction. When 5mm collimation is used, the scan range must frequently be broken into two sets in order not to exceed a comfortable breath-hold (25 seconds). In conventional CT, 8- to 10-mm collimation is employed, with no gap between slices.

Interslice Spacing/Reconstruction Interval Reconstruction interval can be defined as the distance from the beginning of one image to the beginning of the next. Generally, it is equal to the slice thickness so that no gap exists between the two images. When slice thickness equals the reconstruction interval, the images are said to be contiguous. As a general rule, contiguous scanning is

5063_Lee_Ch05pp0225-0310 10/14/05 2:08 PM Page 233

Thorax: Techniques and Normal Anatomy

233

B

A Figure 5-8 A: Computed tomography with contrast in an 88-year-old woman with chest pain shows a pulmonary embolism in the left upper lobe (arrow). B: Because the imaging was performed at 25 seconds after contrast injection, the scan was not optimized for evaluation of the patient’s known lung mass (M) or the unsuspected aortic dissection (F, false lumen).

used in thoracic imaging. With a sequential scanner or single-detector helical CT, certain circumstances may warrant noncontiguous imaging. The most common such indication is high-resolution CT (HRCT). HRCT is employed for the detection or characterization of diffuse lung disease. With MDCT using 1 mm collimation, the same data can be reconstructed as contiguous 5-mm images, contiguous 1 mm images or noncontiguous 1mm images reconstructed at 10 mm intervals. As long as the raw data is present, reconstruction interval in helical CT may be changed after the patient has already been scanned. With sequential scanning, reconstruction interval is set prior to scanning. In certain situations, the reconstruction interval may be selected to be less than the slice thickness. This is particularly true when multiplanar or 3D reconstructions will be performed. In these circumstances, an overlap of 50% to 60% will result in optimized longitudinal resolution for coronal and sagittal reconstructions (34). Overlapping reconstructions are quite useful in tracheobronchial CT or CT angiography of the thorax.

Pitch With single-detector helical CT when table feed is equal to collimation (pitch of 1.0), longitudinal scan coverage is limited. Greater longitudinal coverage is achieved with increased pitch (284). An acceptable compromise between longitudinal scan coverage and broadening of the slice sensitivity profile is achieved when a pitch no greater than 2 is used with a 180
interpolation algorithm (35,37). Routine helical imaging in the thorax is performed with pitch values between 1.0 and 1.5, utilizing collimation in the range of 8 to 10 mm and table movement speeds of 8

to 15 mm per second. With a table speed of 15 mm per second, tube rotation of 360 degrees per second and x-ray beam collimation of 10 mm (pitch of 1.5), during a 24 second helical scanning acquisition, longitudinal coverage of 36 cm is achieved. In practice, complete interpolation data is lacking for the initial and last helical rotations, which thus cannot be reconstructed in the conventional axial plane. This results in a shorter, effective longitudinal coverage that is closer to 34 cm in length. Scan coverage in the longitudinal or z-axis is thus directly related to the variables of collimation and pitch and is proscribed by patient breathholding ability and x-ray tube heat capacity. As might be expected, the ability among patients to perform sustained breath-holding varies considerably and requires an individualized assessment. In ambulatory patients and those without significant cardiopulmonary disease, single breath-hold times of 38 to 56 seconds are possible and as many as 7 consecutive 12-second breath-hold sequences may be performed. Hospitalized patients and those with chronic lung and cardiac disease are generally capable of single breath-hold times of 18 to 32 seconds and may perform 4 to 6 consecutive 12-second breathhold sequences (84). For patients who cannot maintain a single breath-hold for the duration of a complete helical scan, slowly exhaling towards the end of the imaging sequence may be acceptable (264). In the era of single detector CT, pitch was much easier to understand as the value of table travel per gantry rotation divided by collimation. Much of the early helical CT work centered on the benefits and disadvantages of certain pitch values (34,35). The advent of MDCT has changed thoughts about pitch. Pitch may be defined as the table in-

5063_Lee_Ch05pp0225-0310 2/17/06 9:43 AM Page 234

234

Chapter 5

crement/collimation of a single detector (detector pitch) or the table increment/collimation of the beam (detector width times the number of detectors; beam pitch) (233). As slice profiles become less dependent on pitch and vendors vary their definitions of pitch, the pitch value becomes less crucial to designing protocols. For chest CT, pitch should be as close to 1 as possible for single detector CT and for MDCT should be selected to allow for coverage of the entire thorax with one comfortable breath-hold.

Field of View Any size field of view (FOV) has a finite number of pixels (usually 512
512). A trade-off exists between the field of view size and pixel resolution. The larger the field of view chosen, the larger the pixel size and, therefore, the lower the spatial resolution. The field size should be adjusted to encompass only the area to be scanned; in this case, the thoracic cavity. Most examinations should be reconstructed with a FOV restricted to the thoracic wall, just external to the ribs. Exceptions might include those patients with melanoma or other potential subcutaneous abnormalities. In the latter circumstances, the FOV should be increased so as to include the entire skin and soft tissues. If deemed necessary, smaller fields of view may be chosen to better assess a particular anatomic region. The FOV should always be restricted to the lungs whenever high resolution images of the pulmonary parenchyma are acquired. With helical CT, the FOV may be changed retrospectively as long as the raw data is still present (Fig. 5-9). This can be quite useful in situations whereby a targeted reconstruction is needed after the fact (evaluating tracheal stenosis

detected on a routine chest CT). Care must be taken not to restrict the FOV too much, as the area outside of the region is still scanned and an overly restricted FOV may exclude meaningful pathology.

Radiation Dose Relatively high radiation doses are applied in CT examinations as compared to other radiographic studies (178). Although CT represents one-tenth of all x-ray based medical examinations, it is responsible for greater than two thirds of all medical radiation (125). Dose is a particularly important consideration in exposure to the young female breast and in pediatric patients. The dose values within a specific segment are determined by factors such as voltage, current, scan time, scan field, rotation angle, filtration, collimation, section thickness and spacing. Typical multiple scan average doses for body examinations are in the 10 to 40 mGy range (231). It has been demonstrated in conventional CT that a twofold reduction in tube current (from 400 mAs) with an analagous reduction in radiation dose, produces no significant change in subjective image quality or in detection of mediastinal or lung abnormalities for patients of average weight (170) . Because abnormalities in chest CT are so often highcontrast lesions (i.e., lung consolidation against an air background or lymph nodes within mediastinal fat), relatively low radiation dose scans can be employed without sacrificing sensitivity. In thoracic imaging, particularly with regard to young women, one should not increase the mA solely to improve the appearance of the image (i.e., to decrease noise). Special situations in chest CT that may ne-

A

B Figure 5-9 A 38-year-old woman with saphenous coronary artery bypass graft. A: Initial image with 500-mm field of view shows the patent saphenous graft (arrow). B: Targeted reconstruction using the same data more clearly shows the calcification within the graft (arrow).

5063_Lee_Ch05pp0225-0310 2/17/06 9:43 AM Page 235

Thorax: Techniques and Normal Anatomy

cessitate an increased dose through increasing mA are rare. One such situation, for example, would be the evaluation of an obese individual with a solitary pulmonary nodule suspected to be a hamartoma. If fat can be identified with certainty by CT, the nodule could be confidently categorized as benign, whereas without such a definitive finding, the lesion should probably be biopsied or removed. If initial images are not diagnostic, it would be appropriate to increase the mA and target the lesion with a reduced field of view, optimizing the image and allowing for the best opportunity of demonstrating a benign lesion. Recent work with z-axis tube modulation in some of the newer MDCT scanners shows promise as an effective way of limiting CT dose in the chest without compromising image quality (118). Another potential strategy for dose reduction is the use of breast shields, which may reduce breast soft tissue dose without affecting image quality (76).

Computed Tomography Parameters for Viewing Window Settings Among the various contents of the thorax, there is a great range of density values in Hounsfield units (HU), spanning the dense osseous structures (600 HU) to the lung parenchyma (700 HU), which is almost entirely comprised of air. As the wide range of densities recorded by CT exceed that which can be displayed on a single photographic image, unique window settings must be employed to best view lung parenchyma, mediastinal soft tissue, thoracic wall and skeletal structures. As a general rule, at least two window settings are needed in thoracic CT. One setting is used to optimally display the lung parenchyma, and another for the soft tissues of the chest wall and mediastinum. Under certain circumstances, a third set of window settings optimized for osseous detail may be necessary, although the wide window widths used for the lungs almost always adequately display the bony structures. Window setting display can vary depending on personal preference, but should nevertheless follow certain general guidelines. Narrow window widths are rarely needed in the thorax because of the high natural contrast present. If accurate size measurements are sought, the window level should be placed midpoint between the average CT number of the structure to be measured and the average CT number of the surrounding tissue (19,141). As an example, the size of a pulmonary nodule having an attenuation value of 100 HU, surrounded by lung with an average value of 700 HU will be most accurately determined at a window level of 300 HU. Window settings optimized for the lung tend to underestimate the diameter of the bronchial lumen. If precise measurements are desired, a window mean of 150 HU is the most accurate; bronchial wall thickness is best determined using a mean window setting of 450 HU (278).

235

Specific attenuation values are diagnostically important in the performance of chest CT in only limited clinical situations. Most determinations can be made in a relative manner, by comparing abnormal regions, such as pleural collections, with the attenuation of adjacent muscle, to determine if the collection is near water (notably lower than muscle attenuation), or a more complicated collection, such as pus or blood (attenuation close to or higher than muscle). Precise documentation of fat attenuation is of major significance in two lesions: the pulmonary nodule (hamartoma) and focal lung consolidation (lipoid pneumonia).

CT Intravenous Contrast Administration The thorax has the highest intrinsic contrast of any body part. The pulmonary vessels and ribs have markedly different densities than the adjacent lung parenchyma. In the majority of adults, the mediastinal vessels and lymph nodes are embedded in sufficient amounts of surrounding fat such that they can be readily distinguished. Although no specific patient preparation is necessary for thoracic CT imaging, when the use of intravenous contrast is anticipated, some prior restriction of oral intake is preferred to decrease the risk of aspiration if vomiting does occur. When the use of intravenous contrast is indicated, a venous catheter of sufficient caliber to accommodate the desired injection rate should be placed. Ideally, this should be accomplished before the patient enters the CT imaging suite, allowing for the most efficient use of scanner time. The use of intravenous contrast material becomes necessary in the following situations: (a) insufficient mediastinal fat making identification of normal structures difficult; (b) difficulty differentiating an aneurysm or vascular mass from a normal vascular structure or variant; (c) when the particular enhancement characteristics of a mass are desired (e.g., evaluation of a complex pleuroparenchymal abnormality); and (d) the evaluation of a vascular structure (pulmonary embolism, coronary stenosis, aortic dissection) (Fig. 5-10). Assessment of the lung parenchyma, including isolated masses not adjacent to the mediastinum or pleura, almost never requires intravenous contrast. Precise measurement of the degree of contrast enhancement at CT may be helpful in distinguishing a malignant nodule from a benign lesion (260), somewhat analogous to the degree of increased fluoro-deoxyglucose (FDG) metabolic activity in positron emission tomography (PET) imaging (208). These issues are discussed in Chapter 7. The timing of the contrast material injection is critical for optimal vascular enhancement. A programmable volume-flow rate injector, or “power” injector, is indispensable for administering intravenous contrast in a reliable and reproducible fashion. A type 18- or 20-gauge needle with a flexible cannula should be introduced into a medially directed antecubital vein. The antecubital vein is the preferred location of venous access as drainage is directly into the basilic system and superior vena cava, facilitating the integrity of the

5063_Lee_Ch05pp0225-0310 2/17/06 9:43 AM Page 236

236

Chapter 5

A

B Figure 5-10 A 33-year-old man with acute Stanford type A aortic dissection. A: Noncontrast enhanced image at the level of the aortic arch is deceptively normal. B: Only after the administration of intravenous contrast medium is the intimomedial flap seen. When aortic dissection is the clinical concern, intravenous contrast is necessary for its exclusion.

bolus. Central catheters may be used with modification of injection rates. Though some authors have suggested using fast injection rates with central lines, this may not always be possible as certain types of central lines cannot handle the injection rates greater than 2.5 mL per second (234). One must be familiar with the type and brand of central line in use prior to subjecting the central catheter to high injection rates. Regardless of vendor, we routinely use 1.5 mL per second injection rates through central lines and 1.2 mL per second through ports without negative effect. More recently, peripherally inserted central catheters (PICCs) and central lines have been developed that are designed for power-injection capability (up to 5 mL per second). With older-generation CT scanners, 150 mL (240 mg/mL iodine) is injected at 2.0 mL per second for 25 sec, then 1.0 mL per second for the remainder of the bolus. The initial portion of the injection, when no images are being acquired, should be monitored by a technologist (or radiologist) with a finger placed over the injection site for direct palpation, to help ensure that no extravasation occurs. The patient is instructed to notify the technologist immediately of local pain or discomfort from contrast extravasation so that the injection can be aborted and tissue necrosis prevented if ionic contrast is used. With newer MDCT scanners, biphasic injections are rarely needed. The faster coverage allows for the uniform delivery of contrast from 2 to 4 mL per second. Frequently, MDCT allows for reduction in volume of contrast, compared with single detector scanners. Rarely is greater than 100 mL of contrast needed for a chest CT. Scanning is begun after 40 seconds postinitiation of the contrast injection in a CT examination limited to the thorax when 2 mL per second of contrast is used. Scanning usually begins at 25 seconds when 3 mL per second contrast is used. In a busy clinical practice, this technique of a fixed delay works most of the time. A fixed region of interest may be placed over the organ of most interest and semiautomatic bolus timing software may be used as an

alternative to using a fixed delay (Fig. 5-11). This technique requires greater expertise on part of the technologist but is more likely to provide optimal vascular enhancement. At times, because of increased flow from the power injection, retrograde flow of contrast can be seen in collateral branches of the axillary and subclavian vein in the shoulder area. This phenomenon should not be mistaken for venous obstruction (Fig. 5-12). Dynamic scanning is the ability to acquire a series of scans in a short time period. Combined with a rapid bolus injection of intravenous contrast material, it can be performed at a single level after a relatively small bolus (25 to 50 mL), or at sequential levels during the administration of a larger and more sustained bolus (100 mL). Dynamic scanning at a single level after a bolus injection of contrast material provides phasic information and ensures dense opacification of any potential vascular structure under investigation on one of the serial images. Four scans, 8, 15, 22, and 30 seconds after initiation of the bolus injection, usually suffice to visualize the contrast material progressing through the mediastinal arteries and veins. The peak enhancement of a structure reveals whether it has systemic or pulmonary vascular blood supply. Dynamic scanning is more often used in CT examinations of the abdomen, namely in tumor evaluation of the liver, pancreas, and kidney. In our practice, approximately 50% of chest CT examinations are performed in conjunction with a CT scan of another body region, usually the abdomen. Most commonly, the chest CT is combined with an abdominal study, and the most critical organ for contrast administration purposes is the liver. If a fixed delay is used, the scan delay is selected to allow for optimal hepatic enhancement (see Chapter 12). Helical CT has made the task of imaging multiple anatomic areas in a short period of time a real possibility. With MDCT, even longer ranges can be scanned using the same contrast bolus. Routinely, chest, abdomen, and pelvic CT examinations are performed using one bolus and one scan range.

5063_Lee_Ch05pp0225-0310 10/14/05 2:08 PM Page 237

Thorax: Techniques and Normal Anatomy

237

A

B

Figure 5-11 Pulmonary embolism protocol computed tomography using semiautomatic bolus timing. A: A fixed region of interest is placed over the pulmonary artery. B: Low-dose scanning is initiated at a given time (usually 10 seconds), chosen to allow sufficient time for contrast monitoring by the technologist. Once a given threshold is achieved (usually 100 HU), the scanning automatically initiates. C: Contrast-enhancement curve from this examination shows the constant slope of enhancement that is expected in a routine study.

CT SPECIAL TECHNIQUES High-Resolution Computed Tomography of the Thorax High-resolution computed tomography (HRCT) has become established as a useful technique for the detection and characterization of diffuse lung disease. Processes that are nonspecific or inapparent on chest radiographs and conventional CT imaging are shown to advantage by the optimized spatial resolution employed in this scanning method (20,167,185,196,222,224); HRCT has been shown useful in establishing the presence of interstitial lung disease and in characterizing its extent, activity and distribution (Fig. 5-13). HRCT aids in identifying appropriate locations for biopsy, directing intervention towards specific areas in the setting of patchy disease and avoiding regions of end-stage fibrosis. In addition, HRCT may be of value in selected patients to follow the course of disease with time and therapy (56,106). The anatomic detail provided by HRCT imaging sections allows refinement in the localization of disease processes to the small airways, airspaces or alveolar walls. The detection of morphologic abnormalities involving these structures forms the basis for constructing a

C

rational differential diagnosis and occasionally allows for a specific radiographic diagnosis to be suggested. HRCT technique is indicated in a variety of clinical settings (206). These include the evaluation of patients with normal chest radiographs who have either chronic respiratory symptoms or abnormal pulmonary function tests and the further investigation of the chronic and diffusely abnormal chest radiograph in which no diagnosis has been established. HRCT may also play a role in detecting early morphologic changes in patients with suspected diseases such as bronchiectasis, asbestosis and emphysema (87,136). HRCT is based on optimized imaging and reconstruction parameters to demonstrate lung anatomy with the greatest possible spatial resolution. The essential technical features of HRCT include the use of thin collimation, data reconstruction with a high spatial frequency “sharp” or “bone” algorithm, small pixel size derived from the use of a large image matrix and small field of view, and the use of increased kVp or mA (169). The thinnest collimation available on most scanners is 1 to 2 mm. This is closely matched to the size of the anatomic structures of interest within the secondary pulmonary lobule, minimizing volume averaging effects and

5063_Lee_Ch05pp0225-0310 2/17/06 9:44 AM Page 238

238

Chapter 5

A

B

C

D Figure 5-12 Two different patients with contrast-enhanced computed tomography examinations using fast injection rates (4mL/s). A, B: Normal examination showing that normal enhancing veins from a right arm injection. The visualization of multiple venous tributaries should not be confused with collateral vessels recruited from a central venous obstruction. Frequently, multiple venous tributaries including dural collaterals (arrow in A) are opacified as a result of the rapid injection rate with the patient’s arms elevated above the head. C, D: Another patient with a left arm injection and a left brachiocephalic thrombus (T in D). Note the markedly increased number of collaterals compared to the few recruited in A and B.

A

B Figure 5-13 A 58-year-old man with idiopathic interstitial fibrosis (usual interstitial pneumonitis). A: Routine 5-mm-thick image through the lower lobes shows the characteristic honeycombing and irregular septal lines with a basilar and peripheral predominance. Arrow denotes a questionable area of ground glass. B: 1-mm reconstruction through this region (high-resolution computed tomography technique) shows that the area of ground glass (arrow) simply represented volume averaging of irregular septal lines and walls of the honeycomb cysts.

5063_Lee_Ch05pp0225-0310 2/17/06 9:44 AM Page 239

Thorax: Techniques and Normal Anatomy

improving spatial resolution in the longitudinal plane (z axis) (171). Although it is typical to use the narrowest available collimation, the recognition of discriminating features in interstitial lung disease is only slightly different when 1.5 and 3.0 mm collimation are compared (187). Spatial resolution in the transaxial plane is a function of pixel size, which is dependent on the image matrix and field of view (FOV) selected. The largest available matrix size is used, typically 512
512 and the FOV limited to encompass only the lung parenchyma. With a 40 cm FOV and 512
512 matrix, a pixel size of 0.78 mm is achieved. If desired, transaxial plane spatial resolution can be further improved with the use of retrospective reconstruction, targeting a single lung or small portions of lung where disease is thought to be active. A pixel size as small as 0.25 mm can be achieved with an FOV of 13 cm (169,171). The various reconstruction algorithms applied to data acquired in CT scanning can affect the contrast sensitivity, spatial resolution and noise within the resulting image. The high spatial frequency “bone” or “sharp” reconstruction algorithms used in HRCT produce less image smoothing than those used in conventional CT, resulting in increased image sharpness and spatial resolution. Such algorithms have been shown to improve the visibility of small structures, such as vessels and bronchi, when compared to images reconstructed with standard algorithms. The reduced smoothing produced by high spatial frequency reconstruction algorithms also results in increased visible image noise. (169,171,187,292). The use of thin collimation causes increased quantum image noise and decreased contrast sensitivity, as fewer photons are available to reach the CT detector system (171). Compared with standard CT radiographic technique, the use of higher kVp, higher mA or longer scan duration increases the number of photons incident upon the CT detector system. Such changes partially offset the increased noise caused by the use of both thin collimation and high spatial frequency reconstruction algorithms (169). Nevertheless, there is only minimal compromise in diagnostic information when usual “low-dose” technique is compared with “highdose” HRCT techniques (154,164,291). Specifically, lowdose technique has been shown slightly less sensitive in detecting some instances of ground glass opacity and interstitial lines, but otherwise provides similar anatomic information to that of high-dose HRCT technique. Increasing scan duration is generally not employed as a method to decrease image noise, owing to the detrimental effects of motion artifact. Although contrast sensitivity also is worsened by increased image noise, there is little practical adverse impact in the thorax given the intrinsically high contrast of lung parenchyma to surrounding air, in comparison with lesser contrast differences within intra-abdominal structures (e.g., liver lesion) (189,291). Window settings and image formatting for photography are also important variables in HRCT technique. The appearance of general lung density, pleural soft-tissues and

239

apparent size of opaque structures within the lung parenchyma are all greatly affected by the level and width selected for display (278). Level settings from 500 to 750 HU and widths from 1,000 to 2,000 HU may be used to view the lung parenchyma, with specific settings often dictated by individual preference (162,224). The consistent use of at least one routine window setting in all patients is recommended to establish a uniform standard and basis for future comparison. At our institution, a width of 1,800 HU and level of 550 HU is used for routine viewing of the lung parenchyma. Special window settings may be used to accentuate specific disease processes. For example, emphysema may be demonstrated to advantage by the use of low window levels of –800 to –900 HU and narrow widths of 500 HU (Fig. 5-14). HRCT is typically obtained at suspended full inspiration and in the supine position. Prone positioning is occasionally used to differentiate dependent atelectasis from true interstitial lung disease (259). Expiratory HRCT imaging may demonstrate air-trapping in patients with a variety of airway diseases, including asthma, small airway disease, and emphysema. During forced expiration, normal lung increases in density and regions of air-trapping become more visible as relatively lucent areas of lung parenchyma (227,257). A consequence of the maximized spatial resolution in HRCT is the inability to image the entire thoracic volume in a practical manner. With contiguous 1.0-mm collimation, more than 300 images would be required to study the entire chest. HRCT is therefore useful to characterize diffuse lung disease by sampling various levels, rather than as a tool for the comprehensive evaluation of the thorax. Consequently, HRCT may be used as a supplement to prior conventional volumetric CT imaging of the chest, or as a stand-alone technique. HRCT images may be obtained at reproducible preselected levels, such as the aortic arch, tracheal carina and extreme lung bases, at regular intervals ranging from every 10 to 40 mm, or in a concentrated area to investigate or follow more focal disease as identified on plain radiographs, standard chest CT or prior HRCT examinations (167,186,259). HRCT technique may also be modified to address certain specific clinical questions. The highest resolution CT images at a single level must be acquired with conventional scanning techniques. Helical (spiral) scanning technique results in decreased spatial resolution in the longitudinal plane and, in general, is not appropriately combined with HRCT imaging (270). With conventional, sequential scanners and single-detector helical CT, HRCT must be performed as outlined above. The thin (1-mm) images must be acquired separate from the helical scan of the thorax. The advent of MDCT has changed the way HRCT can be performed. With MDCT, if a thin detector collimation is used (e.g., 1 mm), images may be constructed twice as conventional contiguous 5-mm and as thin 1-mm images reconstructed every 10 mm using the same data set. The net result is that the patient may be scanned once and HRCT-equivalent images

5063_Lee_Ch05pp0225-0310 2/17/06 9:44 AM Page 240

240

Chapter 5

A

B Figure 5-14 The effect of window width and center on parenchymal evaluation. A 48-year-old with centrilobular emphysema. A: Routine window of 1,800 HU width and –550 HU center shows the darker regions corresponding to areas of lung destruction. The advantage of this window width is that the soft tissue contrast is adequate to allow for some degree of evaluation. B: A modified window of 500 HU width and –800 HU center more easily identifies regions of low attenuation, corresponding to regions of lung destruction. This window, however, precludes evaluation of soft tissue and osseous structures and makes normal lung appear abnormally high attenuating. Use of consistent window settings allows one to develop experience in identifying normal from abnormal lung.

may be obtained. Initial work has demonstrated that images reconstructed from the helical data remain diagnostic but do have a higher rate of motion artifacts compared with more traditional axial HRCT (130,240). This may be related to the longer scan times of the entire thoracic volume as compared to a single axial slice. Faster scan times

of the 16 and 64 slice scanners should alleviate this problem (Fig. 5-15). The advantage of the use of MDCT is that the entire volume can be reconstructed as 1-mm-thick images without any added radiation to the patient. These images can then be reconstructed in sagittal and coronal planes (Fig. 5-16). This technique, sometimes known as

A

B Figure 5-15 A 50-year-old woman with end-stage lung disease related to scleroderma. A: 1-mm reconstruction from a 4-row volumetric acquisition shows the extensive fibrosis and traction bronchiectasis. Though the images are diagnostic, there is minimal motion artifact in the lingula. B: 1-mm reconstruction from a 64-row volumetric acquisition shows the same findings as in A but no motion artifact (arrow). Faster acquisition times with the newer generation multidetector computed tomography (MDCT) scanners have all but eliminated this potential drawback of MDCT and high-resolution CT scanning.

5063_Lee_Ch05pp0225-0310 2/17/06 9:44 AM Page 241

Thorax: Techniques and Normal Anatomy

volumetric high resolution CT, may allow for better depiction of the vertical distribution of disease and for a more rapid assessment of the lungs than viewing 1-mm axial images alone (197,198,219).

Thin-Section Computed Tomography Imaging Thin-section CT imaging is a technique conceptually distinct from HRCT. Thin-section imaging is primarily employed in the further evaluation of focal lung disease, such as the characterization of pulmonary nodules and detailed evaluation of the airways (148,290). The technique differs from HRCT in that a high spatial frequency reconstruction algorithm usually is not used. The use of thin collimation alone improves longitudinal axis resolution, reducing vol-

241

ume averaging effects while targeted FOV improves XY resolution. The size, internal characteristics, and margins of pulmonary nodules are better characterized. The use of HRCT-type reconstruction algorithms in the evaluation of pulmonary nodules by thin section CT may lead to an erroneous identification of calcification (261). A specific volume is imaged, generally directed toward the further characterization of findings on conventional chest radiographs or standard collimation chest CT images. Complete evaluation requires review of both lung and mediastinal soft tissue window images. When a nodule or area of abnormality is identified, either thin collimation can be used to rescan through the area (on a single-detector scanner) or thin reconstruction can be performed retrospectively if thin collimation was

A

Figure 5-16 A 45-year-old man with lymphangioleiomyomatosis (LAM) and pulmonary hypertension. Volumetric contrast-enhanced study was performed to evaluate the underlying interstitial disease and to exclude chronic pulmonary embolism as a cause for pulmonary hypertension. A: Axial image shows the well-defined cysts characteristic of LAM. B, C: Sagittal and coronal images show the vertical distribution of disease.

B

C

5063_Lee_Ch05pp0225-0310 2/17/06 9:44 AM Page 242

242

Chapter 5

A

B

Figure 5-17 A 78-year-old woman with a left lower lobe spiculated nodule. A: Initial 5-mm lung window shows the spiculated lesion that could represent a bronchogenic neoplasm. B: 5-mm soft-tissue window does not aid in narrowing the differential diagnosis. The mass appears to be soft tissue in attenuation. C: Targeted 1-mm reconstruction clearly shows fat (arrow) within the lesion. The spiculated nature of this fat-containing lesion is diagnostic for lipoid pneumonia.

C

originally used (on a MDCT scanner) (Fig. 5-17). Because of the ability to retrospectively reconstruct through a suspected nodule, we generally aim to scan all potential nodule cases on the MDCT scanners. The thin collimation is used (1 mm) and images are first reconstructed as 5-mm contiguous slices. When a nodule is encountered, the technologist can then be instructed to reconstruct through a given table position using 1-mm contiguous slices using both the soft tissue and the lung kernels.

Pulmonary Nodules The suppression of respiratory misregistration during breathhold helical scanning and volumetric data acquisition allows for improved detection of small pulmonary nodules as compared with conventional CT (58). Lesions at the lung bases, which may be missed on conventional CT imaging owing to excursion of the diaphragm between respiratory cycles, are

more readily detected. In one study in which slice thickness and reconstruction interval were held constant, helical scanning demonstrated a 40% increase over conventional CT in the number of nodules detected (223). Two-Dimensional Multiplanar Reformation and Three-Dimensional Imaging The advent of MDCT has resulted in an explosion of the number of images generated by a single study while simultaneously providing better spatial and temporal resolution in imaging of the thorax. Advances in MDCT and in computer processing have resulted in an increased use of a variety of post-processing techniques in routine clinical work (152). No longer are these applications reserved for special circumstances. The display of complex thoracic anatomy may be enhanced by depiction in formats other than that of standard axial projections and reliance on postprocessing

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 243

Thorax: Techniques and Normal Anatomy

techniques may allow for navigation of a large data set that would not be possible if axial images were viewed alone. For the purpose of generating multiplanar and 3D images, the narrowest possible collimation in used, 1 to 3 mm if possible, with scan coverage focused on the specific volume of interest. Employing a pitch of 1.0 to 1.5 results in higher quality 3D reconstructions than can be obtained from comparable conventional CT scans (115). Overlapping reconstructed helical slices by 25% to 75% suppresses the “stair-stepping” artifact commonly encountered with reformations derived from conventional CT scan data and improves the overall quality of the 3D reconstruction (126,245).

243

Two-dimensional postprocessing reconstructions can easily be performed at the CT scanner or on most picture archiving and communication systems (PACS). These reconstructions in sagittal, coronal and oblique planes can be quite useful in depicting the site of origin of a complex mass or in assessing the relationship of a mass with a curved structure like the diaphragm or heart (Fig. 5-18) (36). MPR images can also be useful in showing the craniocaudual extent of an abnormality or in showing focal abnormalities of structures that travel within the Z-axis, such as the trachea or aorta (Fig. 5-19). A specialized type of MPR, known as curved MPR, can be quite useful in presurgical planning in both the trachea and aorta. In this

B

A

Figure 5-18 A 39-year-old with hemorrhagic bronchogenic cyst. A: Axial image shows the subcarinal mass (M) with contents slightly higher in attenuation than simple fluid. B, C: Coronal multiplanar reconstructions (MPR) show the subcarinal mass (M), which abuts the left atrium (LA). In this case, the MPR images allowed for greater confidence in the preoperative diagnosis of a bronchogenic cyst (see Fig. 5-2).

C

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 244

244

Chapter 5

A

B

C

D

Figure 5-19 A 42-year-old man with tracheobronchopathia osteochondroplastica (TPO). A, B: Axial images show the classic high-attenuating nodularity of the chondrus portion of the trachea. The diagnosis can be made from these axial images. C, D: Coronal and sagittal multiplanar reconstruction images better delineate the craniocaudal extent of disease.

technique, a center line is drawn either manually or automatically within the tubular structure. A 2D reconstruction is, then, made from this central path. Curved MPR is quite accurate in measuring focal dilatations or stenoses of the aorta and trachea and is important in stent-planning and assessing for any subtle change between two CT examinations (Fig. 5-20). It can also easily be performed by a knowledgeable technologist (2). Three-dimensional reformations are commonly performed with maximum intensity projection (MIP), shaded surface display (SSD) techniques and more recently,

volume-rendering (Fig. 5-21) (42). Three-dimensional techniques differ from 2D techniques in that they use all of the volumetric data to generate images in any desired plane. MIPs represent a technique in which the brightest voxels in a given data set are displayed in the desired plane. In other words, a coronal MIP represents a coronal image in which the brightest voxels are displayed for any given point on the X and Z axes. Though the MIP is an easy way to display vascular anatomy, one must be aware of potential pitfalls. Adjacent calcification can obscure the enhancing vessel in a MIP image and no information on 3D

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 245

Thorax: Techniques and Normal Anatomy

245

B

A

Figure 5-20 A 33-year-old man with remote history of trauma and an aortic pseudoaneurysm at the aortic isthmus. A: Axial image shows the post-traumatic pseudoaneurysm (PA). B, C: Oblique and axial curved multiplanar reconstruction images allow for accurate measurement of the pseudoaneurysm for presurgical planning.

relationship is preserved in the MIP reconstruction. Because of reconstruction artifacts, all MIPs should be viewed in conjunction with 2D source images. SSD images are generated by the setting of threshold values, where voxels in the data volume above a given attenuation are set to white and those below this attenuation value set to black. Perspective and depth perception are created by an imaginary light source, which may be placed at any position. Volume segmenting techniques are applied to the image data prior to MIP and SSD imaging to remove unwanted bone and soft tissue structures (35). SSD images tend to give the most aesthetically pleasing images, especially when color based on attenuation is added, but rarely are these images used as the sole diagnostic imaging technique. The threshold used for generat-

C

ing the images may artificially create stenoses in vascular imaging. Also, little information is provided on the internal architecture of a vessel (Fig. 5-22). More recently, volume rendering has gained wide acceptance as an alternative postprocessing technique that may enable one to view large data sets without continually referring to the axials. In volume rendering the entire volume is viewed with added weighting to the plane of interest. The result is that perspective between two overlapping structures is maintained. Images may be viewed in any plane and can be quite helpful in presurgical planning. In the thorax, volume-rendered images may also be displayed as virtual bronchography and virtual bronchoscopy images that can be useful in characterizing tracheobronchial disease (Fig. 5-23) (152).

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 246

246

Chapter 5

A

B

C

D

Figure 5-21 A 75-year-old male with chest pain and hilar mass on

E

chest radiograph. A: Noncontrast study shows displacement of intimal calcification suspicious for a focal dissection. B: Post–intravenous contrast image shows an interomedial dissection flap. Because of the short craniocaudal extent, a penetrating atherosclerotic ulcer was favored. Incidentally noted is a left superior vena cava (arrow). C, D: Oblique multiplanar reconstruction (MPR) image better delineates the short craniocaudal extension of the focal dissection. Note that this represents a two-dimensional (2D) technique. This obliquecoronal image is 3 mm thick and centered at the cross-hairs in C. No information out of this plane is included in the reconstruction. E, F: Curved MPR derived from the sagittal image (line in E). The curved MPR represents a 2D technique in which the vessel is straightened out so that accurate and reproducible measurements may be obtained.

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 247

Thorax: Techniques and Normal Anatomy

F

H

247

G

I

Figure 5-21 (continued) G, H: Shaded surface display technique with the bones segmented. In G, the opacity map was selected so that the bones could not be seen, and in H the map was selected so that the bones were minimally visualized. Note that though the shaded surface gives a dramatic three-dimensional (3D) representation with a virtual light source, little information about the internal architecture of this focal outpouching is seen. Rarely is this technique used for diagnostic purposes. I, J, K: Coronal maximum intensity projection (MIPs) with varying thicknesses: I  7 mm, J  17 mm, and K  50 mm. The advantage of the MIP is that the brightest voxels in any given plane are displayed. Recently, sliding-slab MIPs with varying thicknesses have been employed to depict vascular anatomy. Note the ease of seeing the flap (black arrow in I). As MIP thickness increases, the flap (a dark structure) is hidden by adjacent enhancing vasculature.

J

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 248

248

Chapter 5

K

L

Figure 5-21 (continued ) Conversely, the left superior vena cava (white arrow in I, J, K) is better depicted in its entire course from brachiocephalic vein to coronary sinus on the thick images. This case emphasizes the need to view MIPs in varying thicknesses. One of the major drawbacks of MIP imaging is the lack of 3D perspective when two enhancing structures overlap. In this example, the left superior vena cava may be anterior or posterior to the aortic arch. L, M: Volume-rendered coronal images with varying thicknesses: L  15 mm and M  30 mm. Volume rendering represents inclusion of all of the data in a given volume with weighting to the plane of viewing. The net effect is that the added information is present in the background. The net result is that perspective on crossing enhancing structures is preserved.

M

Helical CT Angiography The rapid longitudinal scan coverage achievable with MDCT technique has allowed for shorter overall scan times and reductions in the volume and concentration of intravenous contrast material used in thoracic CT imaging. MDCT scan times in the thorax of less than 20 seconds have replaced the 2- to 3-minute acquisition times typical of conventional CT. This is achieved without compromise in image quality, as the length of time required for sustained opacification of vascular structures is reduced (110). Helical images of the thorax obtained with 60 mL of 60% iodinated contrast material were judged to have superior vascular contrast enhancement to that of conventional CT images obtained with 120 mL of 60% contrast material (59). Routinely, high-quality CT angiographic images can be obtained with 100 cc of 60% iodinated contrast material on the 4-row and 16-row scanners. With the newer

64-row MDCT scanners, high-quality imaging can be obtained using only 75 mL of contrast. Although an empiric 15-second scan delay may be used to achieve adequate vascular enhancement in the thorax, vessel opacification may be optimized in individual patients by the use of a test bolus technique or by semi-automated bolus tracking techniques (see Fig. 5-11). For the test bolus technique, 10 to 15 mL of contrast material is injected at the anticipated rate and dynamic sequential imaging performed at a level of interest, such as the aortic arch or main pulmonary artery. The time to peak vascular enhancement can thus be refined in individual patients and the appropriate scan delay applied to the diagnostic CT angiogram (266). More recently, semiautomated bolus tracking techniques have gained acceptance to determine the scan delay. In this technique, a fixed region-of-interest is placed over the vessel of interest and low dose scanning is performed beginning at a fixed time. This low-dose

5063_Lee_Ch05pp0225-0310 2/17/06 9:45 AM Page 249

Thorax: Techniques and Normal Anatomy

249

A

B Figure 5-22 A: A 58-year-old man with atypical chest pain. Coronary computed tomography angiogram was requested to evaluate for anomalous coronary artery origin or atherosclerotic disease. Shaded surface display (SSD) image shows the irregular left anterior descending artery (LAD) (white arrow). Though the SSD provides an accurate three-dimensional depiction, little information is given on the degree of vessel narrowing. B: Multiplanar reconstructions at orthogonal planes more clearly show the heavy atherosclerotic disease of the LAD (white arrow).

dynamic scanning continues until a fixed threshold is reached at which point the scanning commences. Use of this technique, which has a variety of vendor-specific terms, usually results in more consistent, homogeneous enhancement than test-bolus or fixed scan delay. Its major drawback is the need for technical expertise in its administration. Once learned, it is fairly easy to master (40). From a single helical scan, vascular structures can be reformatted in any plane or rendered in three dimensions, providing a simplified display of even the most complex vascular anatomy (15,110,193). In particular, evaluation of thoracic aortic disease and detection of pulmonary arterial thromboembolism have been greatly advanced by helical CT scanning technique (60,254). Optimal imaging of the thoracic aorta is performed by determination of the appropriate scan delay, rapid bolus injection of 120 mL of 60% contrast material at 3 to 4 mL per second, thin collimation of 3 mm and table incrementation of 6 mm per second. Two-dimensional multiplanar reformations and 3D renderings are created from overlapping images reconstructed at 2 mm intervals. Imaging strategies such as these have been shown to be highly sensitive and specific in the detection and characterization of thoracic aortic aneurysms and aortic dissection, equivalent to competing techniques of magnetic resonance imaging and transesophageal echocardiography (60,254,287). Applied to the imaging of the pulmonary arteries, similar helical CT angiographic techniques reliably demonstrate emboli in second through fourth division pulmonary arteries. Coupled with multiplanar reconstruction, contrast enhanced

helical CT has come to play an important role in the clinical evaluation of suspected pulmonary embolism (283). Variations in scan technique specific to the detection of pulmonary embolism include calculation of scan delays optimized for pulmonary arterial enhancement and scanning cranially from the diaphragm toward the main pulmonary artery, making best use of limited breath-holding capacity during imaging of the lung bases, where embolic phenomena are most common (57,97,221,225). Helical scanning has also been shown useful in the detection and characterization of pulmonary arteriovenous malformations (AVM) (Fig. 5-24). Using 3D imaging and SSDs derived from 2- or 5-mm-thick sections at a pitch of 1.0, helical scanning reliably identified AVMs without the use of contrast enhancement (218,281).

Airway Imaging Limitations in the evaluation of the airways are imposed by imaging strictly in the axial plane. The thin-section multiplanar and three-dimensional images readily generated from volumetric helical CT data have proven useful in the more comprehensive study of the tracheobronchial tree (149,158). Endobronchial lesions, tracheobronchial stenosis and horizontal webs are well defined. In lung transplantation patients, anastomotic strictures and bronchial dehiscence may be assessed (217). To generate high quality reformatted images, narrow collimation of 2 to 3 mm, pitch of 1.0 to 2.0 and overlapping reconstruction intervals of 1 to 2 mm are selected. Because of its speed of scanning, MDCT allows for the entire tracheobronchial tree to be imaged in

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 250

250

Chapter 5

B

A

C

Figure 5-23 A 53-year-old woman with stridor from prior intubation injury. A: Axial image clearly shows the area of tracheal stenosis. B: Volume-rendered virtual bronchoscopy with caudal perspective shows the area of stenosis (white arrow). These images can be useful in communicating the abnormality to the referring bronchoscopist. C: Virtual bronchographic image clearly shows the stenosis. These images are effective in delineating the entire trachea in one coronal image.

less than 20 seconds or in less time than a breath-hold. Using this type of technique, the origin and course of segmental bronchi may be followed. Imaging in this manner may also allow assessment of airways beyond stenosis that cannot be traversed bronchoscopically (128). In addition to the 2D reconstructions and 3D techniques discussed previously, minimum intensity projection images (minIP) may also be useful in delineating airway anatomy and pathology. The inverse of MIP renderings, in minIP displays, the minimum value of imaginary rays are projected through the acquired data volume and mapped to a gray scale image. As with MIPS, MinIPs are complementary to the axial images and volume-rendered images of the airway (Fig. 5-25).

The Thoracic MR Examination General For all MR imaging of the thorax, either a T1-weighted sequence (short TR, short TE) or axial single-shot (SS) fast spin echo is the initial sequence obtained. Usually a turbo spin echo T1 image or SS fast spin-echo image is obtained because of its speed over conventional spin-echo techniques and its equivalent depiction of anatomy. This is true not only for imaging of chest wall and mediastinal soft tissue structures, but also for imaging the heart and great vessels. Because of a high signal-to-noise ratio and lower motion sensitivity in comparison to T2-weighted images (long TR, long TE), T1-weighted images and SS fast

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 251

Thorax: Techniques and Normal Anatomy

251

A

B Figure 5-24 A 40-year-old woman with brain abscess and suspected pulmonary arteriovenous malformation (AVM). A: A standard pulmonary embolism computed tomography scan was performed after the administration of 75 mL of 68% iodinated contrast, with bolus-tracking. B: Shaded surface display image nicely shows the feeding artery (A) and the draining vein (V). Patient went on to coil embolization.

spin echo are excellent for depiction of anatomy. This is especially true in the mediastinum where the bright signal of fat provides excellent contrast between intermediate signal soft tissue structures, such as lymph nodes, and the black signal void of vascular structures. Because of its sensitivity to the increased water content of tissue, T2-weighted imaging will help further characterize pathologic soft tissue findings. Turbo-spin-echo T2 techniques are also routinely employed in order to decrease the time of scanning. Intravascular gadolinium chelates (described below) have been used with T1-weighted images in order to determine the extent of chest wall or mediastinal tumor invasion, to study the extent of inflammatory or infectious lesions, or to perform magnetic resonance angiography (MRA) (28). With newer scanners, 3D fat-suppressed T1 techniques can be routinely employed for thoracic imaging with gadolinium. These rapid techniques can easily image the entire chest in one breath-hold (31). An advantage of MR over CT is its direct multiplanar imaging capabilities, the avoidance of iodinated contrast material and the lack of ionizing radiation. The small bore of the MR scanner may be problematic in large patients or in patients with claustrophobia. Other contraindications to MR include pacemakers and certain metallic foreign bodies

Motion There are a number of challenges to MR imaging of the thorax. Two of the greatest challenges to overcome are artifacts caused by both respiratory and cardiac motion.

Respiratory Gating The simplest method of eliminating respiratory motion artifact is to suspend respiration entirely via breath-hold. While breath-hold techniques are frequently used, not all patients are able to hold their breath long enough to acquire adequate images (84). This has prompted the use of both respiratory gating and respiratory compensation techniques (13,103,157). Respiratory compensation (also known as Exorcist, ROPE, and COPE) is performed through reordered phase encoding. Throughout the respiratory cycle anterior chest wall motion is monitored with a pressure transducer that surrounds the patient’s chest. The phase encoding steps are then reordered. This reordering of the phase-encoding steps decreases the intensity of respiratory motion artifacts and changes the location of the motion artifacts within the data set. This technique has a potential advantage over respiratory gating in that data acquisition time is not increased. However, the signal averaging involved results in marked resolution loss and obscures fine detail. In addition, the practical aspects of implementing this complicated real-time technique have rendered it less useful. As faster scanning techniques have come into routine clinical practice, the need for such complicated scanning techniques has been reduced. Generally, faster sequences will create higher-quality images than these respiratory compensated techniques. In distinction, MR imaging with respiratory gating is a simple and practical approach to reducing respiratory motion effects. Data are collected during continuous breathing,

5063_Lee_Ch05pp0225-0310 3/10/06 2:37 PM Page 252

252

Chapter 5

A

C

but are used for image reconstruction only if they are collected within a reference range (157). Traditionally, the reference for the respiratory motion was obtained by placing a belt containing a displacement transducer around the upper abdomen (68). Recently a navigator spin echo has been used to monitor diaphragmatic motion. The acceptance and rejection of data can then be made either in real time (232) or retrospectively, after data collection (157). The disadvantage to respiratory gating is that it increases imaging time. Cardiac Motion To reduce cardiac motion, cardiac gating is used. Cardiac motion can be monitored by measuring the electrocardiographic (ECG) signal via placement of MR-compatible

B

Figure 5-25 A 46-year-old woman with a right lower lobe carcinoid tumor. A: Axial image shows the tumor, which has an endobronchial component. B: Coronal oblique–multiplanar reconstruction shows the enhancing mass (M). C: Coronal minimum intensity projection images (MinIP) shows the bronchial cut-off much better than the MPR does. The mass (M) is much harder to see, a potential pitfall of the MinIP technique.

electrodes on either the patient’s chest (ventral surface) or back (dorsal surface). It is thought that placement of the electrodes dorsally may reduce artifact caused by lead motion. The leads should not cross over each other or be looped as this could cause the induction of undesired currents and the possibility of surface burns. The R-R interval is measured and image acquisition is triggered by the R-wave.

Coils The two coils most commonly used in thoracic imaging are the standard gradient body coil and the phased-array surface coil. Unlike early surface coils, which did not provide adequate central body signal, phased-array coils provide good central and peripheral imaging, maintain field homogeneity, and significantly improve signal-to-noise above

5063_Lee_Ch05pp0225-0310 2/17/06 9:45 AM Page 253

Thorax: Techniques and Normal Anatomy

that of the standard body coil (73). Smaller, flexible, surface coils also have been designed and are useful for imaging the superior sulcus and brachial plexus. Dedicated shoulder coils also may be useful for imaging this region.

Contrast Agents The contrast agents for thoracic MR imaging most commonly used, as for abdominal MR imaging, are intravenously administered gadolinium chelates. These include gadopentate dimeglumine (Magnevist; Berlex Laboratories, Wayne, NJ), gadoteridol (Prohance; Squibb Diagnostics, Princeton, NJ), and gadodiamide (Omniscan; Sanofi Winthrop Pharmaceuticals, New York, NY). These are all paramagnetic agents which increase signal and have relaxivity rates of about 4.5 msec-1* mmole-1 (211). Paramagnetic agents are administered prior to T1-weighted image acquisition. The dose administered routinely is 0.1 mmole/kg which, as marketed, is approximately 1 mL/10 kg. The one exception to this in the thorax are double-dose dynamic gadolinium-enhanced sequences; these techniques have been described as an excellent method of demonstrating pathology of the aorta and great vessels (see later) (211). Still at the research stage are intravascular contrast agents (156), which bind to intravascular proteins and, thus, remain in the cardiovascular system longer than conventional MR contrast agents, increasing the time of vessel signal enhancement. Although gadolinium-based contrast agents are considered relatively safe (195), some adverse reactions have been reported (243). As done prior to the administration of iodinated contrast, all patients should be questioned about the history of drug allergy prior to gadolinium administration. Specific Uses Aorta and Great Vessels. Magnetic resonance imaging is an

excellent modality for the study of the aorta and great vessels, and has become an important method of assessing dissection, aneurysms, pseudoaneurysms and congenital anomalies such as coarctation or vascular rings (50,131,194). Double inversion recovery single-shot spin-echo imaging helps create rapid black blood images. This is a “black blood” sequence, providing excellent contrast with the bright signal of mediastinal fat (Fig. 5-26). This sequence is usually obtained in at least the transaxial orientation, but also should be obtained in a second imaging plane. The second imaging plane may be in either the oblique sagittal or the coronal orientation. The oblique sagittal imaging plane lays out the aorta centrally (giving the aorta a “candycane” appearance) and is most useful for evaluating coarctation, or extent of aortic dissection. Standard imaging of the aorta includes ECG-gated SE sequences, plus bright blood gradient echo (gradient-recalled acquisition in the steady state [GRASS]; fast imaging with steady state precession [FISP]; fast low-angle shot [FLASH]) cine sequences (see Fig. 5-26). These are often obtained in the oblique sagittal imaging plane and at various transaxial levels (particularly levels with a questionable finding such as a possi-

253

ble dissection flap). Occasionally phase contrast images may be useful in assessing the direction of flowing blood. A newer method for MR imaging of the aorta and great vessels is dynamic double-dose gadolinium-enhanced 3D imaging (211). At our institution, a 3D Fourier-transform spoiled gradient echo sequence (TR 25, TE 6.9, flip angle 40 degrees) is obtained in the oblique sagittal orientation, first without contrast and then with a test dose of 2 mL of gadolinium. This contrast is power injected, and the time of peak bolus is determined. The actual examination is performed with a double dose of the gadolinium-based agent (.2 mmol/kg), with the power injection carefully timed to obtain images during peak bolus as determined by the initial test dose (Fig. 5-27). Cardiac. Standard cardiac imaging, likewise, should also be performed in at least two planes. Again, usually one imaging plane is transaxial and the second is either in a sagittal or coronal orientation. As in the aorta, black-blood spin-echo sequences are usually the first obtained and are excellent for assessment of anatomic morphology. Newer faster single-shot spin-echo (HASTE) black blood sequences (TR 1000, TE 43, flip angle 180 degrees) (252) are used frequently at our institution, especially in the imaging of congenital heart disease in children, not only because of the excellent quality of the images, but also because of the rapidity of acquisition. Although the sequence has been designed for breath-hold imaging, because of its rapidity, we have been able to image without breath-hold and still obtain excellent results. Furthermore, the rapidity of acquisition appears to decrease artifact secondary to cardiac motion. The addition of preparatory radio-frequency (RF) pulses suppresses unwanted signal from flowing blood. Both the standard SE sequences and HASTE sequence are cardiac–gated. Another rapid technique that can be used to image the heart is 3D gradient recalled echo (GRE) imaging and the newer true fast imaging with steady-state precession (True FISP). The GRE images can be performed with double inversion pulses to create a black blood technique, but also can be performed without inversion pulses to create a bright-blood sequence (102). The True FISP images create high-quality bright blood images that, when combined with cardiac gating, allow for exquisite anatomic detail (Fig. 5-28). Cardiac MR is usually performed to assess congenital disease, but newer cine 2D GRE sequences can demonstrate flow patterns suggestive of valvular stenosis and regurgitation, and both cine and saturation-tagged sequences (spatial modulation of magnetization [SPAMM]; delays alternating with nutations for tailored excitation [DANTE]) have opened up opportunities for assessment of left ventricular function (210,241,289). (See Chapter 9.)

Magnetic Resonance Artifacts in the Chest Although respiratory and cardiac motion artifacts, along with methods of suppressing them, were described earlier, some specific artifactual findings may be found in the

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 254

254

Chapter 5

A

B

D

C Figure 5-26

A 38-year-old woman with a history of aortic coarctation repair. A, B: Single-shot spin-echo axial and coronal images can be scanned quickly (less than 15 seconds) and allow for accurate depiction of anatomy. Arrow in B denotes coarctation repair site. C, D: True FISP axial and oblique-sagittal images allow for assessment of flow and can be gated to cardiac motion, allowing for cine imaging to be performed. Arrow in D indicates repair site without any dephasing or evidence of coarctation recurrence.

chest. “Ghost” or pulsatility artifacts occur in the phaseencoding direction and occasionally simulate lesions within the chest. This phenomenon can be caused not only by the pulsatility of flowing blood, but also by pulsatile cerebrospinal fluid or periodic cardiac or respiratory motion. Entry slice phenomenon, also known as “flow-related enhancement,” occurs in black blood SE sequences and is caused by fresh unsaturated blood flowing into a volume of tissue. As a consequence, the blood in the vessel of the

affected slice is bright rather than black. This usually occurs at the ends of a multisection acquisition (69). Care should be taken not to mistake this finding for slow blood flow or intraluminal clot. A key is the periodicity of the finding. When present, it usually occurs at the end of every group of slices. Susceptibility artifact is the principle reason why magnetic resonance imaging is unsuited to assessment of lung parenchyma. The multiple air-tissue interfaces within the lung parenchyma lead to signal loss

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 255

Thorax: Techniques and Normal Anatomy

A

255

B Figure 5-27 A 22-year-old baseball pitcher with thoracic outlet syndrome. A: Maximum intensity projection magnetic resonance imaging angiography images using 20 mL gadolinium with the arms down do not show any abnormality. B: With the arms up, an additional 20 mL of gadolinium show a focal stenosis in the left subclavian artery indicative of thoracic outlet syndrome.

from intravoxel phase dispersion and decreased field homogeneity. This is most pronounced with gradient echo imaging, but is a general problem with all conventional MR imaging. Wraparound artifact, although not specific to the chest, may be a problem if the object size exceeds the field of view (FOV). This may be a problem when imaging large patients, or when imaging more focal areas, such as the brachial plexus. The phenomenon is usually more severe in the phase-encoding direction. The easiest way to correct this problem is to increase the FOV; however, this option may not be practical because it will decrease spatial resolution. Switching the phase- and frequencyencoding directions will not eliminate the artifact but may place the artifact in a region of the image unnecessary for diagnosis. Other possibilities that will decrease wraparound artifact include the use of a surface coil or the application of saturation pulses outside of the FOV. In addition, most commercial scanners are equipped with “nophase wrap” which is a phase-oversampling software (69). Chemical shift artifact occurs at interfaces between fat and water in the frequency-encoding direction because of the difference in fat and water resonance frequencies (220 Hz at 1.5 T). When both fat and water molecules are included in the same voxel, the signal from the fat molecules shifts to a second voxel in the frequency-encoding direction. In the chest, this is of significance when attempting to accurately measure the size of lymph nodes or other soft tissue structures surrounded by mediastinal fat. Chemical shift misregistration can be reduced by increasing receiver bandwidth, increasing in-plane spatial resolution or

decreasing slice thickness. In addition, it is less pronounced on T1-weighted images in comparison to T2 (179). CT Artifacts Motion is the main source of artifacts in thoracic imaging. In addition, scans through the level of the shoulders may be compromised by artifacts related to beam hardening that occurs when a substantial amount of bone is included in the scan volume. Though somewhat counter-intuitive, a single long breath-hold as used for helical CT scanning, is more easily accomplished in elderly patients and those with marginal lung function, than multiple repetitive for shorter breathholds. Adequate coaching by the technologist or nurseassistant and pre-scan hyperventilation, makes a single extended breath-hold possible for most patients. An artifact related to short scan time acquisition, less than 2 seconds, occurring with both conventional and helial scanning, can lead to potential false positive diagnosis of aortic dissection. The halo effect (250) results in a lowattenuation region adjacent to a vessel, most notably the aorta, simulating the false channel of aortic dissection. This phenomenon is secondary to the pulsation of the vessel which, when rapidly attenuated by the CT x-ray beam, results in superimposition of scan data in the contracted state mismatched with scan data from the more expanded aorta. Certain software packets are now available that segment the acquisition of scan data such that the halo effect does not occur.

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 256

256

Chapter 5

A

B

C

D Figure 5-28 A 51-year-old man with left ventricle thrombus. A, B: Axial and coronal black blood images showing the thrombus (C) within the left ventricle. C, D: Axial and oblique True FISP bright blood images confirm the clot (C). Arrow shows the normal mitral valve in D. With gating, normal anatomy is nicely shown with this technique.

NORMAL THORACIC ANATOMY AND ANATOMICAL VARIANTS CT superbly demonstrates virtually all anatomic structures within the thorax. This is because striking density differences exist between fat, muscle, skeletal structures, and lung parenchyma. Normal fat, which has a low characteristic attenuation value that ranges from 80 to 120 HU, is present in sufficient amounts in most adults to clearly depict most structures of the mediastinum and the chest wall. MRI provides the same morphologic distinctions on as CT but is based on magnetic tissue signal intensity differences rather than on x-ray absorption. Mediastinal and

hilar vascular structures are easily visible on black blood images, because the moving blood in their lumina produces no signal, in comparison to the higher signal intensity of other mediastinal structures. Structures that are air-filled, such as the trachea or bronchi, also are devoid of signal because of low proton density. Muscle, mediastinal lymph nodes, and the collapsed esophagus are of intermediate signal intensity. The bones show a characteristic high signal intensity in the marrow owing to the abundance of fat, whereas the cortical regions are of low signal intensity owing to low proton density. As is true of the entirety of radiology, accurate identification of abnormalities depends on a detailed knowledge

5063_Lee_Ch05pp0225-0310 2/17/06 9:45 AM Page 257

Thorax: Techniques and Normal Anatomy

of normal anatomy as well as an awareness of the wide range of anatomic variants that can exist. Normal anatomy is discussed here in three subsections. First, a brief introduction to thoracic anatomy based on nine representative transaxial images of the chest will be presented. Then, specific structures and regions within the thorax, such as the trachea and the azygoesophageal recess, are discussed in further detail. Finally, the last subsection describes commonly encountered normal anatomical variants in the thorax that might create confusion.

Thoracic Anatomy Although somewhat arbitrary, a series of nine basic levels provides an orderly demonstration of the major structures in the lower neck and chest and their anatomic relationships. These levels, in cephalocaudad order, have been named after the most consistent or dominant anatomic structure: 1. 2. 3. 4. 5. 6. 7. 8. 9.

Thoracic inlet and sternoclavicular joint Crossing left brachiocephalic vein Aortic arch Aortopulmonary window Left pulmonary artery Main and right pulmonary artery Left atrium Cardiac ventricles Retrocrural space

257

Common deviations from these most typical CT patterns are addressed in the section on normal variants.

Thoracic Inlet and Sternoclavicular Junction (Figs. 5-29 and 5-30) These CT planes run characteristically parallel to the clavicles and at a slightly lower level—the first rib. At this level, a number of anatomical structures are displayed on CT or MR without superimposition (269,293). At about the level of the sternal notch, the two lobes of the thyroid gland are typically seen on either side of the trachea and may at times surround this airway structure. The trachea is typically midline and may be in close proximity to the adjacent vertebral body. The esophagus, the other major mediastinal structure at this level, is separated from the trachea by the tracheoesophageal membrane. It usually lies slightly to the left of the midline and frequently contains air but usually is not distended. Five major mediastinal vessels are usually noted anterior and lateral to the trachea. These include the three major branches of the aortic arch (the brachiocephalic, left common carotid, and left subclavian arteries) and the two brachiocephalic veins. The brachiocephalic (innominate) veins, located just posterior to the clavicular heads, are formed near the first ribs at a point where the subclavian veins join the internal jugular veins. The brachiocephalic artery, the largest vessel, lies directly in front of or just to the right of the trachea. The exact position of the bifurcation of the brachiocephalic artery is variable depending on its length and degree of tortuosity,

B

A Figure 5-29 A, B: Contrast-enhanced computed tomography scan above the suprasternal notch and at the level of the thyroid gland demonstrates the jugular veins (jv) just lateral to the enhancing thyroid gland (T). Just posterior to the veins, the common carotid arteries are visible (arrows). The vertebral arteries are seen just lateral to the vertebral body (straight arrow). White T, trachea; Pmj, pectoralis major muscle. Pn, pectoralis minor muscle. Curved arrows: subclavian arteries. H, clavicular heads; SH, sternohyoid muscle.

5063_Lee_Ch05pp0225-0310 2/17/06 9:46 AM Page 258

258

Chapter 5

A

B

Figure 5-30 Sternoclavicular junction. A, B: Through the heads of the clavicles (cl), several major vessels are noted surrounding the trachea (T). The brachiocephalic veins (BV) lie immediately posterior to the clavicular heads. The right brachiocephalic artery has divided into the carotid (c) and subclavian (s) branches. The left carotid artery (c) and the left subclavian artery (s), which commonly indents the left upper lobe, also are seen. Gas within the sternoclavicular joint is common with the arms positioned above the head. Pmj, pectoralis major muscle; Pn, pectoralis minor muscle; arrow, sternothyroid muscle; e, esophagus. sa: subclavian artery, sv: subclavian vein. C: Through the manubrium (M) and its articulation with the head of the clavicles (CL), five major vessels are seen surrounding the trachea (T). The brachiocephalic artery (BC) is the largest of these vessels and lies just anterior to the trachea before it divides into the right common carotid and right subclavian branches. The left common carotid (c) and the left subclavian artery (s) are seen to the left of the trachea. The brachiocephalic veins lie behind the clavicular heads (cl). Note the partly distended esophagus (arrow) behind and slightly to the left of the trachea.

C

which are usually related to the patient’s age. Cephalad to the bifurcation, six major mediastinal vessels will be present, as the right subclavian and right common carotid arteries are seen as discrete structures. The left common carotid artery lies to the left and slightly posterolateral to the brachiocephalic artery; it has the smallest diameter of the three major arterial branches that arise from the aortic arch. The left subclavian artery, which traverses from a more anterior position at its origin to a relatively posterior location, commonly comes in contact with the mediastinal pleural reflection of the left upper lobe and sometimes indents it in a convex fashion. The right brachiocephalic vein usually abuts the right upper lobe.

The subclavian arteries and veins exit and enter the mediastinum by crossing over the first rib, behind the proximal portions of the clavicles. Several anatomic variations can frequently be seen in the origin of the great vessels from the arch (see Figs. 5-77A, 5-79, 5-80). The subclavian vein commonly can be followed as it crosses the first rib to become the axillary vein where it lies behind the pectoralis muscles. The axillary vein lies anterior and the axillary artery posterior to the anterior scalene muscle. The subclavian arteries, however, are much more difficult to trace because of sharper degrees of angulation, because these vessels arch over the apex of the lung slightly cranial and posterior to

5063_Lee_Ch05pp0225-0310 2/17/06 9:46 AM Page 259

Thorax: Techniques and Normal Anatomy

259

the subclavian veins. The subclavian and axillary arteries are accompanied by branches of the brachial plexus, which are rarely visible on CT because of their small size. Although the great vessels generally can be confidently recognized by noting their characteristic configurations on scans, tortuosity and/or ectasia of these vessels may cause confusion, similar to that encountered with conventional radiographs (237). Intimal calcification on CT may clarify the vascular nature of these structures. Alternatively, the use of intravenous contrast media enhancement almost always resolves the problem. Thin collimation and multiplanar reconstructions can be helpful in difficult cases.

and can simulate a lung mass in noncontrast images (213). Anatomic sections at this level may occasionally contain the arch of the azygos vein as it crosses over the right upper lobe bronchus to join the posterior wall of the superior vena cava (Fig. 5-33). In about 90% of the population, the azygos arch is positioned more caudally at the aortopulmonary level (253). The anterior mediastinum in front of the aortic arch typically has a triangular configuration, with the apex pointing anteriorly. In older adults, the anterior mediastinum is composed almost entirely of fat, mainly the residual of normal thymus tissue, though its parenchymal elements generally are not readily apparent.

Crossing Left Brachiocephalic Vein (Fig. 5-31)

Aortopulmonary Window (Fig. 5-34)

This level usually corresponds to the junction between the manubrium and body of the sternum. This level typically contains two brachiocephalic veins anterior to three arteries usually positioned just anterior to the trachea. The right brachiocephalic vein has a nearly vertical orientation throughout its length and is therefore seen in cross-section on transaxial images. The left brachiocephalic vein has a longer and more variable route. It is most often oriented horizontally as it crosses from left to right in the anterior mediastinum to join the left brachiocephalic vein to form the superior vena cava. There is also considerable variability in the size of the left brachiocephalic vein. Mild distention may be a normal variant, although enlargement should alert the surgeon for the presence of venous hypertension or increased flow (see Fig. 5-77A). Intravenous contrast administration in the left medial antecubital vein allows maximal enhancement of the left brachiocephalic vein. The internal mammary arteries and veins also frequently are well visualized (Figs. 5-32A and B), although asymmetry in their size is not unusual, or they may be absent if harvested for bypass surgery (see Fig. 5-32D). Normal thymic veins also may be demonstrated.

Anatomically, this area comprises the region between the aortic arch cephalad and the left pulmonary artery as its caudal margin (111). The aortopulmonary window region contains the distal trachea, variable amounts of mediastinal fat just beneath to the arch medial to the descending aorta and above the left pulmonary artery, and a few small lymph nodes (see Fig. 5-32D). The cephalocaudal dimensions of the aortopulmonary window are variable, and the cephalad portion of the left pulmonary artery may lie just underneath the aortic arch and may fill this potential space. Careful assessment of suspected pathology in this area may require thinly collimated scans with contrast enhancement. The esophagus may be displaced to the left into the aortopulmonary window, especially in the elderly as a result of adhesions between it and a tortuous atherosclerotic descending aorta. Air or contrast within its lumen or contiguity with the esophagus on other scan levels should permit distinction on CT. This can be a potential pitfall on MRI, because the air is not as easily recognized and an abnormal mass may be simulated. The range of normal diameters for the ascending and descending aorta for adults is rather broad. The ascending aorta is always larger than the descending portion at the same level. The ratio can vary from 2.2 in young adults down to 1.1 in older individuals (9). The azygos vein arch is most commonly located at this level; its characteristic course can be traced from its initial prevertebral location as it courses anteriorly (Fig. 5-33A and B) in intimate contact with the right lateral wall of the trachea to enter the posterior aspect of the superior vena cava. The area medial to the junction of the azygos arch and the superior vena cava (sometimes called the paratracheal retrocaval space) usually contains some lymph nodes (often referred to as the azygos nodes), up to 10 mm in diameter, in addition to fat and some connective tissue (Figs. 5-34C and 5-35A) (238). The clinical importance of this region derives from the critical role of these paratracheal (azygos) lymph nodes, which drain the bronchopulmonary (hilar) and subcarinal lymph nodes efferently and are afferent links with the more cephalic middle mediastinal paratracheal nodes. These nodes are commonly involved and enlarged in both inflammatory and neoplastic conditions.

Aortic Arch (Fig. 5-32) The aortic arch usually arises paralleling the superior border of the right second costal cartilage, and after an oblique course, it extends posteriorly and to the left, becoming the descending aorta approximately at the level of the fourth thoracic vertebra. The anterior portion of the arch lies in front of the trachea and is intimately related to the anteromedial aspect of the superior vena cava. The midportion of the arch typically lies just to the left of and causes slight indentation of the trachea. Its most posterior portion, at the junction of the aortic arch and the descending aorta, is just lateral to the esophagus. In the elderly population with advancing arteriosclerosis, the aortic arch often has a cephaloposterior course, with the three great arteries arising sequentially more anteriorly than in younger individuals (see Fig. 5-32C). The brachiocephalic artery tends to originate first at the most caudal level, followed by the left common carotid artery, and finally by the left subclavian artery in the most cephaloposterior region. The top of an ectatic aortic arch may be partially contained within one image

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 260

260

Chapter 5

B

A

D

C Figure 5-31 A, B: Crossing left brachiocephalic vein. Scan demonstrates the left brachiocephalic vein (BV) moving across the anterior mediastinum from left to right to join the right brachiocephalic vein (R) to subsequently form the superior vena cava at a lower level. T: trachea, esophagus (arrow); BC: brachiocephalic artery; c: left carotid artery; s: left subclavian artery; CL: clavicular heads; M: manubrium. C, D: Half-Fourier acquisition single-shot turbo spin-echo (HASTE MR) images in the axial plane. C: At a comparable level to Figure 5-31A demonstrating the same structures. Note internal mammary arteries and veins (arrows). D: Sagittal HASTE MR images demonstrates the left brachiocephalic vein (arrow) sandwiched between the aorta posteriorly and the sternum in front. la: left atrium; ra: right atrium; p: right pulmonary artery.

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 261

Thorax: Techniques and Normal Anatomy

261

A

B

C

D Figure 5-32 A: Top of aortic arch (AA). LV: left brachiocephalic vein R: right brachiocephalic vein. Note internal mammary vein (arrow) and artery (short arrows). B: Mid-aortic arch (AA). Note azygous vein (V) draining into the back of the superior vena cava (C). Note internal mammary artery (arrow) and vein medially (arrowhead ). C: Elongated aortic arch (A) projecting posterior to the left subclavian artery (s). c: left common carotid artery. B: right brachiocephalic artery. BCV: left brachiocephalic vein. T: trachea. D: Reflux of intravenous contrast material from the superior vena cava into azygous vein (v). This azygous arch is located slightly more caudal than the one shown in Figure 5-34B. Note left internal mammary artery bypass graft clip (arrow). Additional clips (arrowheads) are seen in area of harvested mammary vessels. Compare to patent vessels in right side. Multiple small lymph nodes are seen in the aortopulmonary windows.

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 262

262

Chapter 5

B

A Figure 5-33 A: Sagittal half-Fourier acquisition single-shot turbo spin-echo magnetic resonance (HASTE MR) to the right of the midline demonstrates the azygous arch (arrow) draining into the posterior aspect of the superior vena cava (C). B: Coronal HASTE MR illustrating the azygous vein (arrow) seen traveling anteriorly to meet the superior vena cava. A: aorta; T: trachea; P: main pulmonary artery; LA: left atrium.

Left Pulmonary Artery The posterior extension of the main pulmonary artery, the left pulmonary artery lies just to the left and lateral to the carina as it crosses over the left main stem bronchus (see Fig. 5-35). The left upper lobe pulmonary vein is seen anterior before moving medial to the pulmonary artery prior to draining to the upper portion of the left atrium. Typically, the origin of the right upper lobe bronchus from the right main stem bronchus is seen at this same level, as is the right upper lobe pulmonary artery and accompanying bronchus (apical segment). On rare occasion, the main and left pulmonary artery are positioned at an unusually high level relative to the aortic arch and can mimic a mass in the left side of the mediastinum on CT or plain chest radiographs (174). Careful evaluation of sequential scans, and if necessary the intravenous administration of iodinated contrast medium, should allow the proper interpretation to be made.

Main and Right Pulmonary Artery The pulmonary valve marks the beginning of the pulmonary arterial system and frequently is visible on CT (see Fig. 5-46D) or MR. The right pulmonary artery extends posteriorly and to the right from the main pulmonary artery (Fig. 5-36), coursing posterior to the superior vena cava and anterior to the bronchus intermedius (see Fig. 5-35B). The intrapericardial portion of the right pulmonary artery should be visible in all patients, and normally measures 12 to 15 mm in diameter in adults (201). The normal diameter of the main pulmonary artery usually is about two thirds that of the ascending aorta

and should not exceed 29 mm (147). The main as well as the left pulmonary artery can be enlarged in patients with pulmonary valvular stenosis. The left interlobar pulmonary artery is usually present just posterolateral to the left upper lobe bronchus. The right and left superior pulmonary veins generally are seen respectively just anterior to the lateral aspect of the right pulmonary artery and the left upper lobe bronchus. The medial part of the right lower lobe is seen immediately behind the bronchus intermedius as it insinuates itself into the azygoesophageal recess. Similarly, lung frequently inserts into a small notch between the left interlobar pulmonary artery and the descending aorta.

Left Atrium (Fig. 5-37) Lying between the aortic root and right atrium anteriorly, and the azygos vein, esophagus and descending aorta posteriorly, the left atrium has anteroposterior measurements through its midportion of about 3 to 4.5 cm (101). The inferior pulmonary veins course into its posterolateral aspect. Congenital variations in the number and size of the pulmonary veins are relatively frequent and can be crucial when mapping prior to radiofrequency catheter ablation (122,168). The aortic root lies to the right of and posterior to the main pulmonary artery and right ventricular outflow tract. The pulmonary valve can be visible on thinner sections as a triangular membrane and typically lies at the level of the right atrial appendage and is more cephalad than the aortic valve. The aortic sinuses may give the aortic root an ovoid configuration, where it is slightly larger in diameter

5063_Lee_Ch05pp0225-0310 2/17/06 9:46 AM Page 263

Thorax: Techniques and Normal Anatomy

263

B

A

Figure 5-34 Aortopulmonary window. A: Just underneath the aortic arch, the calcified ductus ligament can be seen (arrow). AA, ascending aorta; DA, descending aorta; E, esophagus; T, trachea. Upper lobe pulmonary arteries and veins are well opacified bilaterally. B: At a level just slightly caudal to that shown in A, several small lymph nodes are seen (arrow). Note fatty atrophy of the thymus (short arrows) in this 53-year-old woman. A, azygous; T, trachea; E, esophagus; C: Just slightly caudal to B, the top of the left pulmonary artery can be seen (P). Note calcified mediastinal lymph node (arrow).

than the more cephalad tubular ascending aorta. The proximal coronary arteries are often visible in the mediastinal fat as they originate from the aortic sinuses. Current CT techniques combine high speed and spatial resolution with sophisticated electrocardiographic synchronization and robustness of use. Application of these modalities for evaluation of coronary artery disease is a topic of active current research (114,239). (See Chapter 9.) Subepicardial fat may be abundant within and demarcates the interatrial and atrioventricular grooves. The azygoesophageal recess is also seen at this level and is almost always concave in adults; it becomes deeper and more medial with increasing age and the presence of emphysema (160). In children and in adults with a narrow anteroposterior diameter of the thorax, the normal deep lung concavity of the recess

C

may display a medial convexity because of bulging of the esophagus (204). The appearance of the normal azygoesophageal recess in children and its differentiation of pathologic conditions is important to recognize, because, in general, mediastinal abnormalities that affect the azygoesophageal recess are nonspecific (176). The left atrial appendage is visible just behind vertical portion of the main pulmonary artery and above the proximal left coronary artery. A small ridge of tissue mimicking an incomplete septation is frequently visible between the left atrial appendage and the left lower lobe vein (see Fig. 5-37B). The three sinuses of Valsalva (right, left, and noncoronary) are visible as subtle areas of outpouchings just above the aortic valve (see Fig. 5-37C).

5063_Lee_Ch05pp0225-0310 2/17/06 9:46 AM Page 264

264

Chapter 5

B

A Figure 5-35 A: Left pulmonary artery. The left pulmonary artery (L) is seen arching above the left main stem bronchus. White arrow demonstrates the origin of the left upper lobe branch. Black arrow: Retroaortic pericardial recess. Arrowhead, small pre-carinal lymph nodes. B: At a slightly lower level, a fatty-involuted thymus is seen in the anterior mediastinum (arrows). L, left pulmonary artery.

B

A Figure 5-36 A: Right pulmonary artery. The proximal right pulmonary artery (RPA) is seen arching behind the ascending aorta and superior vena cava (C). V, right upper lobe pulmonary vein. B: Left main bronchus. A, left upper lobe pulmonary artery; LPA, left pulmonary artery; Arrow, left upper lobe pulmonary vein; Curved arrow: anterior segment branch of the right upper lobe pulmonary artery. V, Right upper lobe pulmonary vein. B: At a slightly lower level, the left upper lobe pulmonary vein is seen coursing toward the left atrium. V, right upper lobe pulmonary vein; RPA, right pulmonary artery. A, left atrial appendage; L, left lower lobe artery.

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 265

Thorax: Techniques and Normal Anatomy

265

A

B

C

D Figure 5-37 A: Left atrium: The left atrial appendage (LAA) is seen posterior to and to the left of the pulmonary outflow tract (P) A, aortic root; V, right upper lobe pulmonary vein; S, superior vena cava. Arrow points to right atrial appendage. B: Slightly lower level in another patient and at a level just below the left atrial appendage, the left coronary artery and its proximal branches are visible (arrow). LA, left atrium; V, left lower lobe vein; V, right upper middle lobe vein confluence; A, right lower lobe artery. Curved arrow, right middle lobe artery. C: At a slightly lower level in another patient, the right coronary artery (arrow) is seen arising between the right atrial appendage (RAA) and right ventricular outflow tract (OT). C, superior vena cava; A, aortic root. D: Just below B, the right atrial appendage (RAA) is visible as it partly wraps around the aortic root. V, right lower lobe pulmonary vein. Black arrow, right middle lobe veins. LV, top of left ventricular outflow tract. White arrow, small pericardial recess behind right lower lobe vein (v). RV, right ventricle.

Cardiac Ventricles (Fig. 5-38) The right ventricle, both anterior to and to the right of the left ventricle, represents a significant portion of the heart’s anterior surface. It is separated from the left ventricle by the obliquely oriented interventricular septum. The interventricular septum is readily visible when intravenous contrast is administered, or in noncontrast images when the

patient is severely anemic. Contrast medium is usually required to define the myocardium, however. Normally, the left ventricular myocardium is about three times thicker than the right ventricular wall (see Fig. 5-38B). There is sometimes a shallow impression on the anterior surface of the heart over the area of the interventricular groove; the left anterior descending coronary artery may be visible

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 266

266

Chapter 5

A

B

C

D

Figure 5-38 A: Cardiac ventricular level. The two mitral valve

E

leaflets (arrows) are seen at this level. V: lower lobe pulmonary veins. B: CT slice at midventricular level demonstrating all four cardiac chambers. Transvenous pacemaker (arrows) causes metallic artifacts in right atrium and right ventricle. C: Just inferior to B, the coronary sinus (CS) is demonstrated as it drains into the right atrium (RA). Note moderator band (long arrow) as it extends from right ventricular free wall toward the ventricular septum. Filling defect in left ventricular chamber (short arrow) represents papillary muscle. D: At the base of the ventricles, the distal right coronary artery (arrow) is seen coursing in the right atrioventricular groove. IVC, inferior vena cava; Curved arrows, normal pericardium. E: Black blood half-Fourier acquisition single-shot turbo spin-echo magnetic resonance image at the same ventricular level in another patient.

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 267

Thorax: Techniques and Normal Anatomy

267

A

B

C

D Figure 5-39 Retrocrural space. A: The right and left diaphragmatic crura (arrows) are well depicted at this level. Azygous and hemiazygous veins are seen posterior to the descending aorta (A). B, C, D: Coronal black-blood half-Fourier acquisition single-shot turbo spin-echo magnetic resonance images demonstrating from anterior to posterior, the diaphragmatic crura (arrows) bilaterally. The azygous vein (curved arrow) is seen on D as it incorporates the hemiazygous (lower arrowheads) and accessory hemiazygous (upper arrowhead) veins. A, descending aorta; LA, left atrium.

within the fat in the indentation. The pericardium overlying the anterolateral aspect of the heart is almost always seen in adults, Caudally, the horizontal portion of the coronary sinus, the main venous drainage of the cardiac muscle, appears as a tongue-like structure in the inferior portion of the left atrioventricular groove and medial to the inferior vena cava (see Fig. 5-38C and D). Occasionally,

on more cephalad scans the vertical portion of the coronary sinus may be seen along the left posterobasal aspect of the left atrium (175).

Retrocrural Space (Fig. 5-39) The posterior mediastinum extends caudally beyond the level of posterior pulmonary sulci, enclosed by the diaphragmatic

5063_Lee_Ch05pp0225-0310 2/17/06 9:47 AM Page 268

268

Chapter 5

crura. The retrocrural space, at the level of the aponeurotic hiatus in the diaphragm, allows the passage of the esophagus and the descending aorta and also includes fat, the azygos and hemiazygos veins, the thoracic duct, and associated lymph nodes (246). The normal nodes in this area should not exceed 6 mm in cross-sectional diameter (43,44,66).

Specific Anatomic Structures Brachial Plexus (Fig. 5-40) Brachial plexopathy can be caused by a variety of pathologic conditions, such as trauma and intrinsic or extrinsic tumors. CT has important limitations in the direct evaluation of the brachial plexus. These include the difficulty or inability in the transaxial plane to follow the plexus in its entirety, as well as beam hardening artifacts due to adjacent bones. MRI is the method of choice for evaluating patients with a nontraumatic brachial plexopathy (282). Tumors most likely to involve the brachial plexus include invasion by superior sulcus (Pancoast tumor) bronchogenic tumors and metastatic breast carcinoma (140). Radiation therapy for either of these tumors or for treatment of lymphoma can also cause brachial plexopathy (166). Also to be considered as an cause of brachial plexopathy are various forms of neuritis from diverse causes. The brachial plexus arises from nerve roots C5 to T1, which then form the upper, middle and lower trunks. From the trunks arise the anterior and posterior divisions, then the lateral, posterior, and medial cords and, last, the radial, median, and ulnar branches (30). Specifically, the anatomic path of the brachial plexus is as follows: the nerve roots run from the spinal cord through the intervertebral neuroforamina. The ventral rami of the roots run anterior to the transverse process and posterior to the vertebral artery. It is

these ventral rami that give rise to the brachial plexus. Next, the trunks run between the anterior and middle scalene muscles and superior to the subclavian artery. The divisions of the brachial plexus run through the supraclavicular triangle, which is formed by the sternocleidomastoid muscle, the inferior belly of the omohyoid muscle, the middle third of the clavicles, and the trapezius muscles (30). Through this triangle the divisions continue superior to the subclavian artery. After the lateral border of the first rib, the divisions become the cords that enter the axilla. The cords then form the radial, median, and ulnar nerve branches. When performing an MR examination of the brachial plexus, a narrow slice thickness should be used on all sequences. T1-weighted images should be obtained in at least two planes, and a third is often helpful. Because of its high signal-to-noise ratio, T1-weighted images best define the anatomy of the brachial plexus. The sagittal plane is often the most diagnostic, demonstrating the brachial plexus in cross-section (282). Sagittal images are prescribed from a transaxial or coronal scout and should include the neuroforamina and all nerve roots from C-5 to T-1. Although unilateral sagittal images dedicated to the affected side will demonstrate anatomic detail, either a coronal or a transaxial T1-weighted set covering both the affected and nonaffected side will permit direct comparison for symmetry. T2weighted images help to better characterize pathology and should be selected in the imaging planes that best demonstrate the disease. In selected cases, particularly if neoplastic invasion is in question, intravenous gadolinium-enhanced T1-weighted images may be helpful. The accuracy of tumor detection with MR imaging can be limited by the presence of diffusely infiltrating tumor. PET, with the use of quantitative tissue characterization, can be used to detect the extent of disease with potentially more specific tumor identification than is possible with MR imaging and may reveal the presence of unsuspected metastases outside the axilla. Correlation of PET data with the anatomic information derived from MR imaging resulted in data that were helpful in determining the approach to both local and systemic therapy in patients suspected of having recurrent or advanced axillary disease (108).

Chest Wall

Figure 5-40 Magnetic resonance image of brachial plexus. Sagittal T1-weighted image of the normal brachial plexus. Visible are the peripheral nerve branches (arrows), the subclavian vein (v) and artery (a), the subclavian muscle (s), and the clavicle (c).

Bilateral symmetry is relatively characteristic and facilitates the radiologic analysis of the chest wall. However, the shoulder girdle is quite mobile and its symmetry and appearance on CT varies dramatically with the position of the upper extremities. Most cross-sectional illustrations in anatomy textbooks are based on adduction of the arms alongside the body. Although such positioning is used for MR, it may create linear or “streak” beam hardening artifacts on CT. Optimal CT imaging in most patients will require that the arms be abducted in external rotation and elevated alongside the head. In this position the scapulohumeral joint, the spine of the scapula, and the coracoid process are seen at about the T2 to T3

5063_Lee_Ch05pp0225-0310 2/17/06 9:47 AM Page 269

Thorax: Techniques and Normal Anatomy

269

A

Figure 5-41 A: The wide manubrium of the sternum (m) can have an indistinct posterior margin as compared with the sharply defined anterior cortical boundary. B: The narrower body (B) of the sternum usually has sharp, cortical margins in its entire periphery.

B

vertebral body level. The sternoclavicular joints are usually seen at about the T4 level. The three sections of the sternum, its articulations with the clavicles, and the contiguous soft tissues can be clearly depicted on transaxial CT sections (63,99,109,256). The manubrium is the widest part of the sternum (Fig. 5-41A and B) and forms the anterior boundary of the superior mediastinum. Because of some angulation, part of its posterior cortical margin normally is unsharp and irregular, and even the anterior margin may be indistinct. Such an appearance could simulate a destructive lesion. If desired, the gantry can be angled so that it is perpendicular to the

manubrium in an effort to improve cortical definition. The superior border of the manubrium has a rounded defect, the jugular (sternal) notch. Just cephalad, the proximal portions of the clavicles have a steep obliquity and appear in cross-section as elongated or oval structures. Caudal to the jugular notch, there is an indentation (clavicular notch) on the posterolateral aspects of the manubrium for articulation with the clavicular heads. The costal cartilages of the first ribs articulate with the manubrium anterolaterally. A very prominent first costochondral junction, a well-recognized pitfall on conventional chest radiography, can cause a rounded opacity on a lung window setting CT scan that

5063_Lee_Ch05pp0225-0310 2/17/06 9:47 AM Page 270

270

Chapter 5

A

B Figure 5-42 Pulmonary pseudonodule. A: Partial volume averaging through the caudal portion of the first ribs mimicking a pulmonary nodule (arrow). B: Scan at a slightly higher level shows that the nodule is due to the result of sclerosis of the medial portion of the first costosternal articulation (arrow).

may mimic a pulmonary nodule (Fig. 5-42A and B) (207) or a poorly defined area of parenchymal consolidation. Typically, such confusion arises on wider collimated CT sections approximately 1 cm below the junction of the first rib with its prominent costal cartilage and is a result of averaging of the rib with the surrounding lung. The “lesion” appears intraparenchymal rather than part of the anterior chest wall because of the rapid oblique caudal and anterior sloping of the latter and the protrusion of the rib into the coronal plane of the lung. Close inspection and awareness of the existence of this partial volume averaging effect is usually sufficient for its correct interpretation; however, if there is doubt, thinly collimated scans through the region may be definitive. The more caudal body of the sternum is ovoid or round and usually has sharp cortical margins (see Fig. 5-41B). Its lateral surfaces may be irregular at the various costochondral junctions; soft tissue prominences also may surround these articulations. Similarly, sections through the transition between the manubrium and body of the sternum, and between the body and xyphoid process, typically demonstrate indistinct or sclerotic margins. Fusion anomalies may involve the lower half of the body of the sternum in 4% of the population. Clefts (slits) or wide gaps (foramen) may be seen on CT (256). In pectus excavatum deformities, the sternum is not only posteriorly displaced, but also usually rotated somewhat obliquely. Multidetector row CT with 3D volume rendering has proved worthwhile in musculoskeletal imaging and can also be applied to imaging of the thoracic cage including pectus excavatum deformity. The complex

curving, overlapping anatomy of the shoulder girdle and thoracic cage is well suited to 3D imaging (152). The ribs are essential structures of the osseous thorax and provide information that aids in the interpretation of radiologic images. Techniques for making precise identification of the ribs are useful in detection of rib lesions and localization of lung lesions (146). Ribs curve through sequential transaxial CT images in a predictable manner, allowing identification of rib number (27,134,145). The most straightforward method (134) uses the sternal angle as the primary anatomic landmark, where the high-density oval portion of the sternum without bone marrow attaches to the second costal cartilage. This is usually seen on the aortic arch or aorticopulmonary window scan level. At that transaxial level, ribs 2 to 6 will usually be displayed from anterior to posterior. The sixth rib can then be followed on successive scan levels, as it curves anteriorly. The other ribs (7 to 12) can then be subsequently identified at lower levels. The internal mammary vessels, located about 12 mm lateral to the sternal border, are routinely seen on mediastinal window CT images (see Fig. 32A and B). On lung window settings, these vessels may cause tiny projections into the pulmonary parenchyma (213). The normal adjacent parasternal (internal mammary) lymph nodes are too small to be seen on CT. The cross-sectional appearance of the normal female breast includes a wide spectrum of appearances depending on size, tissue characteristics, and patient position. The parenchyma may be dense and even somewhat mass-like, especially in

5063_Lee_Ch05pp0225-0310 2/17/06 9:47 AM Page 271

Thorax: Techniques and Normal Anatomy

271

B

A Figure 5-43 Chest wall musculature. A: At the level of the manubrium. PMJ, pectoralis major muscle; PN, pectoralis minor; LD, latissimus dorsi and teres major muscles; LT, long head of triceps muscle; Sp, supraspinatus muscle; Sc, scapula; D, deltoid muscle; Is, infraspinatus muscle; SB, subscapularis muscle; Rh: rhomboideus major muscle; Tz, trapezius muscle; E, Erector spinae muscle; I, intercostal muscles. B: Just above the diaphragm: SA, serratus anterior muscle; LD, latissimus dorsi muscle; RA, rectus abdomini muscle.

young women, or quite sparse with mainly fatty tissue in older adults. Normally, the breasts appear symmetric. Symmetry is also useful in the assessment of the axillae. The axillary artery and vein, accompanied by branches of the brachial plexus in a neurovascular sheath, often can be defined on CT as distinct structures (72). The axillary vein begins at the lower border of the teres major muscle and courses cephalad. With the arms positioned over the head, the axillary vein is completely anterior to the artery throughout its course, becoming the subclavian vein at the lateral border of the first rib. Interpectoral lymph nodes, also called Rotter’s nodes, are located between the major and minor pectoralis muscles (Fig. 5-43). Enlargement of these nodes, involved primarily in breast cancer, upper extremity melanoma, and lymphoma, may be seen on CT (113).

Pulmonary Parenchyma Normal lung architecture and appearance as depicted by CT are reasonably constant, though they will vary somewhat depending on the collimation and the phase of respiration. When images are acquired at full lung capacity (end-inspiratory volume), pulmonary vessels will be maximally displayed, improving recognition of small pathologic pulmonary abnormalities. At resting lung volume or in expiration, there is considerable crowding of pulmonary vessels, especially in the dependent portion of the lung.

This phenomenon can be seen in the lung images from an abdominal CT examination, which is performed in expiration. Even with scans done in full inspiration in the supine position, crowding of the vessels posteriorly in the bases and some concomitant dependent atelectasis may be a problem, simulating interstitial disease. By rescanning patients in the prone position, the area is better distended and the confusion clarified (see Fig. 5-7). Normally, vessels will be seen on both thickly and thinly collimated scans virtually all the way to the periphery on the lung images, but there is almost always a small subpleural radiolucent zone, 3 to 5 mm in width, devoid of vessels. The bronchi, in distinction, normally are not seen in the outer 2 to 3 cm of the lung parenchyma. The normal range of pulmonary density generally is between
350 and
650 HU, with the differences less on inspiration than expiration. Mean attenuation values in the posterior third of the lungs may be as much as 100 HU greater than the anterior third; the gradient is reduced, but not abolished, on scans obtained in deep inspiration. When the smaller peripheral pulmonary vessels course horizontally, it is generally possible to trace their origin to the major segmental vessels and verify their nature. If, however, a small vessel courses obliquely or in a cephalocaudal direction, demonstration of the origin (and thus the etiology) of the ovoid or round density usually requires examination of sequential scans (and often, multiplanar reconstruction) to verify that the structure is vascular.

5063_Lee_Ch05pp0225-0310 2/17/06 9:47 AM Page 272

272

Chapter 5

Figure 5-44 Schematic depiction of segmental bronchi. Levels a to g represent key levels and correspond to the levels illustrated in Fig. 5-62. RUL, right upper lobe with its segments, 1R, apical; 2R, posterior; 3R, anterior; RML, right middle lobe with its segments, 4R, lateral; 5R, medial; RLL, right lower lobe with its segments, 6R, superior; 7R, medial basal; 8R, anterior basal; 9R, lateral basal; 10R, posterior basal; LUL, left upper lobe with its segments of the superior division; 1, 2L, apical posterior; 3L, anterior; Ling, lingular division with its segments, 4L, superior; 5L, inferior; LLL, left lower lobe with its segments; 6L, superior; 7, 8L, anteromedial basal; 9L, lateral basal; 10L, posterior basal.

Understanding the anatomy of the secondary lobule is crucial to the recognition and categorization of parenchymal pulmonary abnormalities on HRCT (21). However, the normal HRCT appearance of the secondary lobule has at least two different appearances depending on its location in the lung, whether centrally in the medulla or peripherally in the cortex (280). In addition, and most important, the secondary lobule usually is recognized as a definite structure only when it is abnormal. Normal secondary lobules and even acini may be seen on HRCT if very small fields of view are used or if they are surrounded by abnormal lung (88). In the medulla, the secondary lobule appears roughly hexagonal, whereas in the lung periphery (cortex) it often has a truncated cone shape with the apex pointed centrally. In both regions, the normal secondary lobule has a thin, smooth border. The secondary lobule varies in size from 6 to 25 mm in diameter and can appear polyhydric, prismatic, or the aforementioned truncated pyramidal shape; the centrilobular artery (about 1 mm in diameter) and the bronchiole (about 2 mm in diameter) enter at the central apex. Several perilobular veins are present in the septa, but the septa themselves are less than 1 mm in thickness and are rarely seen normally. This unit, the secondary lobule, generally comprises approximately 5 to 6 acini.

Pulmonary Hila CT images of pulmonary hila (see Figs. 5-35 and 5-36; Fig. 5-45) demonstrate the main bronchi, pulmonary arteries, superior and inferior pulmonary veins, and normal bronchopulmonary (hilar) lymph nodes; the latter are commonly seen on contrast-enhanced helical scans. The

bronchi are the most constant hilar structures and serve as reference points for identification of vessels and lymph nodes. Other structures in the hilum, such as lymphatic vessels, nerve plexi, and areolar tissue, all within a connective tissue envelope, are not defined on MR or CT images (112). An appreciation of the usual relationships among the various anatomic structures (188,270,279) allows recognition of relatively subtle abnormalities in this region, most notably hilar adenopathy.

The Airways Detailed knowledge of bronchial anatomy (see Figs. 5-44 and 5-45) is crucial in pinpointing the exact location of pathologic conditions, particularly when partial lung resection of nodules or tumors is being contemplated (135,192). The bronchi can be used as easily identifiable reference points because they are arranged perpendicular, obliquely, or parallel to the transaxial plane; each particular bronchus has a different, and sometimes variable, appearance, making multiplanar reconstruction (MPR) images frequently helpful. The origin and proximal portion of the horizontally coursing bronchi are routinely imaged on the transaxial CT sections. These include the right upper lobe bronchus (as well as its anterior and posterior segmental branches), the left upper lobe bronchus (and its anterior segment), the middle lobe bronchus (and generally some portion of both the medial and lateral segmental branches), and the superior segments of both lower lobes. Bronchi having a vertical course are seen in cross-section. These include the apical segmental bronchus of the right upper lobe, the

5063_Lee_Ch05pp0225-0310 2/17/06 9:47 AM Page 273

Thorax: Techniques and Normal Anatomy

273

A

B

C

D Figure 5-45 Tracheobronchial anatomy: A: Trachea (T) is seen at the level of the manubrium. The right tracheal wall is rather thin, whereas on the left side, the adjacent left subclavian artery (a) gives it a thicker appearance. B: The trachea has a more triangular appearance at this level just below the aortic arch. Note the apical segment (A) of the right upper lobe bronchus. A similar branch is seen on the contralateral side. C: The right (R) and left (L) central bronchi are seen at this level just below the tracheal bifurcation. Note the anterior (A) and posterior (P) branches of the right upper lobe bronchus. The artery is seen running parallel and medial to the anterior segment bronchus branch. The left upper lobe pulmonary vein (arrowhead) is seen just lateral to the left pulmonary artery. D: Just below the level of C, the anterior (A) and apico-posterior (P) bronchi are visible. I, bronchus intermedius; L, left bronchus.

5063_Lee_Ch05pp0225-0310 2/17/06 9:47 AM Page 274

274

Chapter 5

E

F

G

H Figure 5-45 (continued) E: At a slightly lower level, the bronchus intermedius (arrowhead) is seen behind the right pulmonary artery. The lingular bronchial branch (short arrow) and the superior segment bronchus (arrow) are seen on the left. F: The origin of the right superior segment (short arrow) and right middle lobe bronchus (arrowhead) are noted at a slightly lower level. The left-sided arrow illustrates a lingular bronchial branch (long arrow) as it travels obliquely. G: The right middle lobe bronchus (arrow) is noted as it divides into lateral and medial branches. Curved arrow: left lower lobe bronchus. Arrowhead: lingular branch. H: This computed tomography slice taken at the lung bases illustrates the basal bronchi. 1: posterobasal; 2: lateral basal; 3: anterior basal; 4: medial basal. Each bronchus is accompanied by a corresponding artery. Anterior and medial are branches of the same trunk on the left side.

5063_Lee_Ch05pp0225-0310 2/17/06 9:47 AM Page 275

Thorax: Techniques and Normal Anatomy

apical posterior segmental bronchus of the left upper lobe, the bronchus intermedius, and the proximal portions of both lower lobe bronchi (after the origin of the superior segments). The most difficult bronchi to recognize are those that course obliquely. These are the lingular bronchus (including the superior and inferior segments), as well as the basilar segmental bronchi. The lingular bronchi may at times be visible, depending on the patient’s body habitus and the plane of section. Sometimes, the medial and lateral segments of the middle lobe course with a shallow obliquity. Minor variations in bronchial anatomy are common, but true anomalies are rare. The most cephalad recognizable bronchus is that of the apical segment of the right upper lobe. Characteristic of upper lobe relationships, this bronchus lies immediately lateral to the corresponding branch of the right upper lobe pulmonary artery and just medial to a branch of the right superior pulmonary vein. The right upper lobe bronchus is seen just below the carina and originates more cephalad than the left upper lobe bronchus. The posterior wall of the right upper lobe bronchus is in direct contact with the aerated posterior segment of the right upper lobe. The thickness of this wall generally is quite uniform in caliber and almost always less than 5 mm in thickness. A small focal bulge may be produced by a prominent azygos vein. The anterior and posterior segmental bronchi are usually visible as they bifurcate and course horizontally in the transaxial plane. The bronchus intermedius, which is about 3 cm long, lies directly posterior to the right pulmonary artery and, at a slightly more caudal level, is just medial to the right interlobar pulmonary artery. The entire thin posterior wall of this bronchus is in contact with the aerated superior segment of the right lower lobe in most patients. However, in about 5% of normal individuals, a branch of the pulmonary vein may present as a focal nodularity in the posterior wall of the bronchus intermedius and should not be mistaken for pathology (Fig. 5-46) (133). Typically, pulmonary parenchyma also extend posteromedial to the bronchus intermedius into the azygoesophageal recess. The middle lobe bronchus courses caudally as well as anteriorly. In most patients, the origin of the middle lobe bronchus is demonstrated in the same transaxial plane as the proximal portion of the right lower lobe bronchus (see Fig. 5-45F). The medial and lateral segmental bronchi usually are located slightly more caudal (see Fig 5-45G). Although the superior segmental bronchus of the right lower lobe may arise at the same level as the orifice of the middle lobe bronchus, it can also arise at a slightly more caudal or cephalad level. There is considerable variability in the branching of the right lower lobe basilar segmental bronchi (see Fig. 5-45H). The medial basal segmental bronchus usually lies just anterior to the right inferior pulmonary vein. The anterior, lateral, and posterior basal bronchi may be identifiable because of their positions relative to one another. Characteristic of

275

lower lobe relationships, the basal bronchi lie medial and anterior to the corresponding vertically coursing lower lobe pulmonary arteries and peripherally are slightly lateral to the inferior pulmonary veins. The origin of the left apical posterior segmental bronchus is seen in cross-section at the level of the carina and right upper lobe orifice (see Fig. 5-45D). The left upper lobe bronchus (see Fig. 5-45C) originates at a level more caudal than that of the right upper lobe bronchus; it forms a “sling” over which the left pulmonary artery passes. The orifice of the left upper lobe bronchus is large, and two serial CT sections may include it. The left upper lobe pulmonary artery, as it descends posteriorly, often produces a slight concavity to the posterior wall of the cephalad portion of the bronchus. The aerated superior segment of the left lower lobe abuts the posteromedial wall of the left upper lobe bronchus. The anterior segmental bronchus usually can be identified as the only bronchus arising from the left upper lobe bronchus that courses directly anteriorly in the horizontal plane (see Fig. 5-45D). A section through the caudal portion of the left upper lobe bronchus also often includes the origin of the left lower lobe bronchus. The lingular bronchus arises from the undersurface of the distal left upper lobe bronchus and has an oblique anterior and caudal course (Fig. 5-45E). Serial sections below the orifice of the lingular bronchus show progressively wider separation between the lingular bronchus anteriorly and the left lower lobe bronchus posteriorly. The superior and especially the inferior lingular segmental bronchi are rarely recognized on routinely collimated CT sections. The left superior segment bronchus usually originates at the same level as its right-sided counterpart, although it may originate at a slightly more cephalad level than its opposite mate (see Fig. 5-45F). The left lower lobe basal segmental bronchi generally have a similar anatomic distribution as those on the right side, except that the anterior and medial basal segmental bronchi on the left typically arise as a single structure. Whenever deemed necessary, thin sections through the hilar regions can be helpful in identifying individual bronchial origins. At least the approximate position of most of the bronchopulmonary segments can be recognized by identifying the distribution of the majority of the segmental bronchi (120,205) (Fig. 5-47). The majority of the segments can be confidently predicted and the relative positions of the rest estimated. Thinner-section CT has improved the ability to discern smaller bronchial segments. The hilar vessels are more variable in position than are the bronchi. The truncus anterior or right upper lobe pulmonary artery, the first main branch of the right pulmonary artery, arises within the pericardium and is commonly seen just anterior to the right upper lobe bronchus (see Fig. 5-36A; Fig. 5-48). The posterior branch of the right superior pulmonary vein usually lies within the angle formed by the bifurcation of the right upper lobe bronchus into the anterior

5063_Lee_Ch05pp0225-0310 2/17/06 9:48 AM Page 276

276

Chapter 5

A

B

C

D Figure 5-46 A, B, C, D: Pulmonary vein branch (arrows) coursing behind the bronchus intermedius before draining into left atrium

and posterior segments; the anterior branch generally is located medial to the branches of the truncus anterior. The right pulmonary artery initially crosses in front of the bronchus intermedius, and then at a more caudal level lies alongside its lateral aspect as it becomes the right interlobar pulmonary artery (see Fig. 5-37B), often causing a slightly nodular contour to the hilum at this level. More caudally, the right interlobar pulmonary artery assumes a vertical orientation and is positioned just lateral to the bi-

furcation of the middle and lower lobe bronchial orifices. The right superior pulmonary vein now lies anteromedial to the middle lobe bronchus as it enters the upper portion of the left atrium. Just caudal, the middle lobe pulmonary artery lies between the medial and lateral segmental bronchi of the right middle lobe. The inferior pulmonary veins are oriented relatively horizontal as they enter the lower portion of the left atrium (see Fig. 5-38A).

5063_Lee_Ch05pp0225-0310 2/17/06 9:48 AM Page 277

Thorax: Techniques and Normal Anatomy

277

Figure 5-47 Schematic drawings of segmental lung anatomy at representative levels. A: Aortic arch. B: Left pulmonary artery. C: Right pulmonary artery. D: Cardiac ventricles.

The left upper lobe pulmonary artery and the left superior pulmonary vein have a much less constant anatomic relationship to the left upper lobe bronchus than their counterparts on the right side. The apical posterior bronchus may lie medial or lateral to the branch of the left upper lobe pulmonary artery. The left pulmonary artery itself, first seen cephalad between the left main bronchus and the apical posterior bronchus, generally lies posterolateral to the left superior pulmonary vein. The posterior wall of the left upper lobe bronchus is usually concave where it is indented superiorly and posteriorly by the left pulmonary artery coursing over and around it. The left superior pulmonary vein invariably lies in front of the left upper lobe bronchus as it courses to the left atrium; it frequently has a horizontal course at a slightly more caudal level (see Fig. 5-36B). The left pulmonary artery extends caudally posterior to the left upper lobe bronchus to become the left interlobar pulmonary artery. Pulmonary parenchyma usually can be seen extending into a sulcus between the left interlobar pulmonary artery and the descending thoracic aorta (see Fig. 5-45D). Obliteration of this space just behind the left upper lobe

bronchus (or thickening of the left retrobronchial stripe), similar to the region behind the upper lobe and intermediate bronchus on the right, is pathologic (275). Just caudal to the left upper lobe bronchus, frequently at the level where the bronchus to the superior segment of the left lower lobe originates, the left interlobar pulmonary artery lies in the bifurcation behind the lingular bronchus and anterolateral to the lower lobe bronchus. The segmental lower lobe pulmonary arteries generally lie more toward the periphery of the lung than their corresponding bronchi, whereas the segmental veins are positioned just central to the bronchi. Sometimes, the segmental lower lobe pulmonary arteries may be duplicated or even triplicated; such normal variations may simulate enlarged intersegmental lymph nodes. Several normal areas of notable lobularity in the hila are caused by some of these described vascular relationships. The most marked regions of lobularity in the right hilum occur where the truncus anterior vessel produces a convexity anterior to the right upper lobe bronchus (see Fig. 5-45C), where the right superior pulmonary vein crosses anterolateral to the descending right interlobar

5063_Lee_Ch05pp0225-0310 2/17/06 9:48 AM Page 278

278

Chapter 5

A

B

C

D Figure 5-48 Vascular anatomy of pulmonary hila. A: Contrast-enhanced computed tomography scan depicting the vascular components of the pulmonary hila. AA, ascending aorta; PA, main pulmonary artery. Arrow: right upper lobe vein. V, left upper lobe pulmonary vein. Arrowhead: superior segment right lower lobe pulmonary artery. B: At a slightly lower level. L, left lower lobe pulmonary artery. Arrow: right upper lobe pulmonary vein. The right atrial appendage is visible (A). R: right pulmonary artery. C: The superior vena cava is seen sandwiched between the right atrial appendage and the right upper lobe pulmonary vein (white arrow), and the black arrow demonstrates the proximal left coronary artery. V, left upper lobe vein; D: PV, left and right lower lobe pulmonary veins. LAA, left atrial appendage; RA, right atrium.

pulmonary artery (see Fig. 5-45D) and just caudal to the right middle lobe bronchus where the basilar branches of the right lower lobe pulmonary artery divide. In the left hilum, the most prominent lobulation occurs where the descending left pulmonary artery branches posterior and

caudal to the origin of the left upper lobe bronchus. If these lobulations cause any interpretative problem after viewing serial contiguous scans, contrast-enhanced images should allow confident recognition of vascular structures and eliminate confusion with a pathologic mass.

5063_Lee_Ch05pp0225-0310 2/17/06 9:48 AM Page 279

Thorax: Techniques and Normal Anatomy

The bronchopulmonary (hilar) lymph nodes, normally not seen as discrete structures on noncontrast CT scans, are located adjacent to the bronchi in close relationship to the pulmonary vessels. Cephalad in the hila, the lymph nodes are usually medial to the segmental bronchi and lateral to the pulmonary arteries. Knowledge of hilar lymph node anatomy assists in differentiating lymph nodes from pulmonary embolism. The review of sagittal and coronal reformatted images can help in difficult cases (283). Caudad in the hila, the nodes are situated lateral to the lower lobe bronchi and medial to the pulmonary artery branches. Normal hilar lymph nodes and/or collections of fat are frequently visible on MRI. Collections of tissue, representing both fat and normal-size lymph nodes, large enough to be confused with abnormally enlarged hilar lymph nodes may be visible in three specific locations on MRI: on the right at the level of the bifurcation of the right pulmonary artery, at the origin of the middle lobe bronchus, and on the left at the level of the upper lobe bronchus (277, 279). Generally, lymph nodes and fat are easily distinguished by virtue of their different signal intensities on T1-weighted images. Fat-suppression sequences may be of help in this differentiation if required. Short inversion time inversion-recovery (STIR) turbo spin-echo (TSE) magnetic resonance (MR) imaging has been used for detection of metastases in lymph nodes by using quantitative and qualitative analyses to enable differentiation of lymph nodes with metastasis from those without (200). Bronchial arteries can be visible in good-quality contrast-enhanced scans. Occasionally, they may become enlarged and visible as is the case of patients with right-sided obstructive congenital heart defects or chronic obstructive lung disease. Trachea The trachea is a cartilaginous and membranous tube that extends from the larynx (about the level of the C6 vertebral body) to the carina (about the level of the T5 vertebral body); the extrathoracic length is approximately 2 to 4 cm, and the intrathoracic length about 6 to 9 cm. As it becomes intrathoracic, the trachea lies in the midline or slightly to the right; progressively caudal sections demonstrate a very gradual posterior course. At all levels, the trachea lies anterior or slightly to the right of the esophagus (see Fig. 5-45A and B). The normal tracheal wall is relatively thin and is defined internally by the central air column. The ring cartilages may be irregularly calcified in the elderly, especially in women. There is marked variability in the cross-sectional appearance of the trachea (82). It is generally circular in shape just below the cricoid cartilage and tends to be transversely ovoid near the carina. In between, it may be horse-shoe–shaped; the anterior wall is convex, whereas the posterior wall, lacking cartilage, is flat. The trachea tends to be more uniformly round in children (100). In adults, the mean cross-sectional area of the normal trachea

279

is 272 mm2 (S.D. 33) in men and 194 mm2 (S.D. 35) in women; values less than 190 mm2 in men or 120 mm2 in women should suggest pathology (272). Abnormal variations in the configuration of the trachea are easily defined on CT (70). The most common is the saber sheath trachea (Fig. 5-49), typically associated with advanced chronic obstructive pulmonary disease. In this condition, there is marked narrowing of the width (transverse diameter) of the intrathoracic trachea, to about one half to two thirds of its widened anteroposterior diameter, often down to the level of the carina. The ring cartilages also are frequently densely calcified, particularly in elderly females.

Esophagus When sufficient mediastinal fat is present, the esophagus can be delineated on sequential CT images (see Figs. 5-30B and C and Fig. 5-31A and B). The collapsed esophagus may be confused with a lymph node mass on MRI. The anterior wall of the esophagus generally contacts the posterior aspect of the trachea, the left main stem bronchus, and the left atrium in its caudal descent. Distally, it is bordered by the descending aorta on its left and by the azygos vein on its right. Intraluminal air is a common finding, present in at least 70% of normal patients (see Fig. 5-34; Fig. 5-50). The lumen may appear to be divided into two or three sections by strands of mucus bridging the anterior and posterior walls (see Fig. 5-31B) (213). The apparent thickness of the wall is variable and dependent on the degree of distention. As a general guideline, a wall thickness of 3 mm or more should be considered abnormal (104).

Figure 5-49 Saber sheath configuration of the trachea (T) characterized by increased anteroposterior diameter and narrow transverse diameter. This condition is usually seen in patients with chronic obstructive pulmonary disease.

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 280

280

Chapter 5

A

B

C

D Figure 5-50 A, B, C, D: Normal esophagus partially filled with air (arrows), demonstrating its anatomic relationship at different mediastinal levels.

Thymus (Fig. 5-51) The morphology of the thymus changes drastically with age, and there is a wide variation in its normal size, weight, and consistency. Knowledge of these various appearances is essential for proper CT interpretation (65). The thymus is a bilobed organ; each lobe has a separate fibrous capsule that is connected cephalically to the inferior lobes of the thyroid gland. The bulk of the organ is positioned anterior to the great vessels. The thymus is

relatively largest (with respect to whole body weight) in the neonate and young infant (average weight of 20 g), and increases slightly in size to reach a maximal weight at puberty (average of 30 g). During this period, the thymus consists chiefly of parenchyma, composed mainly of a dense population of lymphocytes separated by thin fibrous septa. Beginning at puberty, a phase of involution takes place, usually over a period of 5 to 15 years. There is a gradual reduction in the lymphocyte population, with

5063_Lee_Ch05pp0225-0310 2/17/06 9:48 AM Page 281

Thorax: Techniques and Normal Anatomy

281

A

B

C

D Figure 5-51 Normal thymic variability with age. A: Normal thymus (arrows) in 18-year-old woman. B: Scanty amount of thymic tissue with fat infiltration is seen on this 50-year-old patient, while complete fat atrophy is seen on C on another patient of the same age. D: Black blood halfFourier acquisition single-shot turbo spin-echo magnetic resonance also demonstrating complete fatty infiltration of thymus.

progressive replacement of the atrophied thymic follicles by fat. After age 40, adipose tissue becomes the dominant constituent of the organ; the parenchymal elements and connective tissue septa become increasingly sparse (see Fig. 5-51B). By the age of 60, the only thymic remnant usually present is fat. The appearance of the thymus on cross-sectional imaging is directly related to these changes (17,74,181). From

birth to puberty, the thymus is seen in the anterior mediastinum, demarcated cephalad by the horizontal portion of the left brachiocephalic vein and caudally by the horizontal portion of the right pulmonary artery. Its largest portion is found in the 3- to 4-cm interval between the aortic arch and the main pulmonary artery. The thymus has a CT density comparable to that of the chest wall musculature in this age group. Its outer contours may be convex

5063_Lee_Ch05pp0225-0310 2/17/06 9:48 AM Page 282

282

Chapter 5

laterally. Although the lobes may appear to merge and assume a triangular (or arrowhead) configuration, they are not completely fused. The thymus often is rotated slightly to the left at the level of the aortic arch, and the junction between the two lobes may be about 2 cm to the left of midline. The left lobe is usually slightly larger than the main caudal extension of the organ. From puberty to 30 years of age, the appearance of the thymus begins to reflect the fatty infiltration that is occurring. Its overall attenuation value has diminished and is always less than that of skeletal muscle. The thymus may be seen as a discrete triangular or bilobed structure; its outer borders are now straight or concave laterally. As emphasized, there is broad variation in this involution. Some organs contain considerable fat in patients under 20, whereas others have little fat even at age 30. Over the age of 30, small islands of soft tissue density commonly are seen within a background of more abundant fat. The residual thymic parenchyma may assume a linear, oval or small round configuration. Never, however, is a focal alteration in the outer contour of the thymus normally seen in adults. Ultimately, in the older patient, only a thin fibrous skeleton remains of the thymus, and the anterior mediastinum is almost totally composed of fat within its remnant (see Fig. 5-51C and D). This fat may be slightly higher in density than that of the subcutaneous fat. Although generally not necessary for accurate recognition of abnormality, the most reliable and meaningful measurements of the thymus relate to its thickness (largest distance across the long axis of each lobe) (Fig. 5-52). With the exception of some neonates, before the age of 20 years, when the thymus is most prominent, 1.8 cm is the normal maximal thickness. Thereafter, 1.3 cm should not be exceeded. The thymus also is visible in all adults on MRI (see Fig. 51D) (61). Distinction between higher-intensity mediastinal fat and the relatively lower-intensity thymus is optimal on T1-weighted images because of the long T1 of the thymus.

The progressive decrease in T1 with advancing age reflects the previously mentioned fatty infiltration. The T2 relaxation times of the thymus do not change with age and overlap with those of fat. The thymus may appear slightly thicker on MRI than on CT in patients over the age of 20 years.

Azygos Venous System The azygos and hemiazygos veins represent continuations into the thorax of the right and left ascending lumbar veins, respectively. The azygos vein traverses the diaphragm through a small opening, usually accompanied by branches of the sympathetic neural chain. It ascends in the right prevertebral area, usually just posterolateral to the esophagus; laterally, it is in contact with the pleural reflections of the right lower lobe where it defines the medial border of the azygoesophageal recess. At the level of the T5 or T6 vertebra, the azygos vein arches toward the right and then anterior over the right mainstem bronchus to drain into the posterior aspect of the superior vena cava (see Figs. 5-32D and 5-33A and B; Fig. 5-53). Sometimes the azygos vein may be interposed between the posterior wall of the right mainstem or upper lobe bronchus and lung in the azygoesophageal recess, simulating a mass or an enlarged lymph node (150). Tracing the course of the azygos vein on contiguous sections generally allows confident identification of this normal venous structure (160,176). On rare occasion, tortuosity of the azygos vein arch may simulate a paratracheal mass (228). Intravenous contrast enhancement should be used to ensure correct interpretation in perplexing cases.

Figure 5-52 Determining thymic size. A: In a bilobed organ, the short axis or thickness (T ) is measured perpendicular to the long axis or width (W). B: When the two lobes are confluent, the thymus is arbitrarily divided in half by a line (x – x1) through the apex of the organ. The thickness (T) can then be measured. (From Baron RL, Lee JKT, Sagel SS, et al. Computed tomography of the normal thymus. Radiology 1982;142:121–125.)

Figure 5-53 Azygous arch (arrows) is seen draining into the posterior portion of the superior vena cava (C).

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 283

Thorax: Techniques and Normal Anatomy

The right superior intercostal vein, which receives flow from the right second through fourth intercostal veins, descends along the right anterolateral vertebral margin, where it may bulge slightly into the contour of the right upper lobe (150). It drains into the azygos vein posteriorly just prior to the formation of the azygos venous arch. The hemiazygos vein on the left side, usually not seen on CT, parallels the posterior aspect of the descending aorta after penetrating the diaphragm through the aortic hiatus. The hemiazygos vein generally crosses the midline prevertebrally, passing posterior to the aorta at about the T8 level, to drain into the azygos vein (Fig. 5-54). The accessory hemiazygos vein, whose caudal course is also posterior to the descending aorta, usually joins the hemiazygos vein just before it crosses the midline (see Fig. 5-39D). The left superior intercostal vein, on occasion seen on cross-sectional imaging, may communicate with the accessory hemiazygos vein at or just above the level of the aortic arch. The left superior intercostal vein then forms a horizontal arch that courses anteriorly along the superolateral border of the aortic arch or just cephalad to join the posterior aspect of the left brachiocephalic vein (14,151,159). Sometimes the close relationship of the left superior intercostal vein to the aortic arch may simulate an aortic dissection on both axial CT

283

and MRI studies. On frontal plain chest roentgenograms (77), the left superior intercostal vein may be seen “on end” just lateral to or above the aortic knob and has been termed the “aortic nipple” (172). This venous pathway may serve as collateral circulation in cases of obstruction.

Mediastinal Lymph Nodes The ability to recognize normal, as well as abnormal, lymph nodes on CT is directly related to the amount of mediastinal fat present. Lymph nodes appear as round or oval-shaped soft tissue or calcific densities within the mediastinum (see Fig. 5-53; Figs. 5-55A and B, 5-57). Normal lymph nodes can be seen in about 90% of adults, especially in the middle mediastinum. Originally a surgical-based classification, (89), a modification (Fig. 5-56; Table 5-1) of the American Thoracic Society Classification, applicable to CT findings, allows dialogue among radiologists, surgeons, and pathologists (91,92). Different nodal groups can be accessed by different methods, depending on their location (Fig. 5-57, Table 5-2). In autopsy dissections, the average number of mediastinal lymph nodes found has been 64; the majority of these nodes are located in the right paratracheal region (levels 2R, 4R ATS) adjacent to the tracheobronchial tree, (216). On occasions,

A

Figure 5-54 A: A 43-year-old woman with a recent history of left innominate vein thrombosis after catheter placement. B: Contrast enhancement of left superior intercostal vein (arrow) after left arm vein injection. C: Contrast material is seen flowing retrogradely into the superior vena cava (C) via the azygos vein (curved arrow).

B

C

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 284

284

Chapter 5

A

B Figure 5-55 Normal mediastinal lymph nodes. A: Slightly enlarged lymph node (arrow) is seen in the pretracheal area just medial to the azygous vein. This is station 10R, also called azygous node, and the number of nodes can range up to three. B: Two calcified lymph nodes (arrows) are seen surrounding the proximal left pulmonary artery (11L).

with both CT and MRI, a closely clustered group of normal-size lymph nodes will appear as one conglomerate enlarged mass. There is a wide reported range, from 6 to 14 mm in diameter, for the maximal size of normal mediastinal lymph nodes (10,86,89). Some of this variation relates to different lymph node regions. Normal nodes rarely exceed 6 mm in diameter around the left brachiocephalic vein, in the supradiaphragmatic region, and in the retrocrural area. The largest size normally encountered is in the right tracheobronchial region, in the aortopulmonary window, and in the subcarinal space, where approximately 90% of normal nodes measure 11 mm in diameter or less. Upper paratracheal lymph nodes are usually smaller than lower paratracheal or tracheobronchial nodes, and right-sided tracheobronchial nodes are larger than those on the left side. Both the short axis and the long axis of mediastinal nodes have been advocated as the more predictable measure of nodal enlargement (volume); 1 cm is considered the upper limit of normal (216). Such measurements are more accurate with right-sided than left-sided lymph nodes. Besides the given limitation that CT cannot detect microscopic architectural alterations (e.g., metastatic neoplasm) in normal-size lymph nodes, it is obvious that no size criterion will allow both a high sensitivity and specificity for predicting abnormal nodal enlargement. For simplistic and pragmatic reasons, generally to enhance sensitivity at the expense of specificity, in our practice any lymph node exceeding 1 cm in diameter, in any dimension, is deemed abnormal and worthy of

consideration for tissue sampling. A mildly enlarged lymph node, containing a central area of fat attenuation, is generally considered a normal variant. (Lymph nodes in the paratracheal and pretracheal areas are accessible to biopsy through a transcervically inserted mediastinoscope [Carlens procedure]). However, sampling of the nodes in the anterior prevascular space, aortopulmonary window, or hilar regions requires a limited parasternal approach (mediastinotomy or Chamberlain procedure).

Pleura and Pericardium Fluorine 18-fluorodeoxyglucose positron emission tomography (FDG-PET) has been reported to be more sensitive than CT in the detection of primary tumor and distant metastases. For example, the accuracy of FDG-PET and CT in detection of primary tumor and metastasis to individual lymph node groups and for nodal staging has been examined. FDG-PET was found to be more sensitive than CT for depicting nodal metastases in patients with squamous cell carcinoma of the esophagus (286). Similar results have been obtained with other types of cancer spread to thoracic lymph node chains (7,153). Pleura The pleura is a double serous membrane that, under normal circumstances, is not visible on CT. The thicker parietal pleura lines the inside the thoracic cavity, being separated from the musculature of the chest wall by a layer of loose areolar tissue. The visceral pleura, which intimately

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 285

Thorax: Techniques and Normal Anatomy

285

Figure 5-56 Lymph node mapping scheme, modified from American Thoracic Society Classification. Nodal stations 6 and 9 are not illustrated because they superimpose other stations on the frontal view. (From Cimmino CV. The esophageal-pleural stripe: an update. Radiology 1981;140:607–613, with permission.)

envelopes the lung, also extends between its lobes to line the major and minor fissures (Fig. 5-58). Delineation of the fissures on CT may be necessary for accurate localization of disease within the pulmonary parenchyma. Similarly, correct identification of loculated fluid within the fissures presupposes knowledge of their normal location and appearance (163). The appearance of the major fissures, oriented relatively vertical (cephalocaudad), is variable depending on the scan collimation and their axis in the individual patient (1,90,165). The major fissures, with wider collimation, appear as somewhat indistinct broad, lucent bands within the pulmonary parenchyma. They appear to be “avascular” because of the diminutive tapering of the pulmonary vessels in the most peripheral portions of the lungs. This is similar in appearance to the normal subpleural radiolucent zone, 3 to 5 mm in width, that is present in the

nondependent portions of the lung, where the peripheral blood vessels are too small to be seen. Less frequently, except on thin sections, the major fissures are seen as discrete lines; this can occur if the vertical axis of the fissure is exactly perpendicular to the plane of the CT section. As emphasized, this requirement can almost always be achieved with narrow collimation (1 to 3 mm) and the major fissures recognized with certainty using this technical variation. A small collection of supradiaphragmatic fat, probably a lateral extension of the pericardial fat pads, is occasionally seen invaginating into the caudal aspect of one or both major interlobar fissures (Fig. 5-59) (81). Similarly, fat from the anterior mediastinum may extend into the anterior junction line between the visceral pleura of the two lungs (Fig. 5-60). The minor fissure, which is relatively parallel to the transaxial CT image, appears as a broad hypovascular area in

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 286

286

Chapter 5

TABLE 5-1

TABLE 5-2

MODIFIED AMERICAN THORACIC SOCIETY CLASSIFICATION OF REGIONAL NODAL STATIONS

LYMPH NODE SAMPLING

Nodal Stations X 2R

2L

4R

4L

5

6 7

8

9 10R

10L

11

14

Method of Sampling

Nodal Stations

Mediastinoscopy

2R, 4R, 4L, 10R 2L nodes adjacent to trachea 7 nodes in anterior subcarinal area Same nodal groups as mediastinoscopy, but 7 nodes more easily accessible 10L and 11 nodes can also be reached 5, 6, and 2L nodes anterior to great vessels Under CT guidance, potentially can reach all nodal groups

Definitions Supraclavicular nodes. Right upper paratracheal nodes. Nodes to the right of the midline of the trachea, between the intersection of the caudal margin of the brachiocephalic artery with the trachea and the apex of the lung or above the level of the aortic arch. Left upper paratracheal nodes. Nodes to the left of the midline of the trachea, between the top of the aortic arch and the apex of the lung. Right lower paratracheal nodes. Nodes to the right of the midline of the trachea, between the cephalic border of the azygos vein and the intersection of the caudal margin of the brachiocephalic artery with the right side of the trachea or the top of the aortic arch. Left lower paratracheal nodes. Nodes to the left of the midline of the trachea, between the top of the aortic arch and the level of the carina, medial to the ligamentum arteriosum. Aortopulmonary nodes. Subaortic and paraaortic nodes, lateral to the ligamentum arteriosum or the aorta or left pulmonary artery, proximal to the first branch of the left pulmonary artery. Anterior mediastinal nodes. Nodes anterior to the ascending aorta or the innominate artery. Subcarinal nodes. Nodes arising caudal to the carina of the trachea but not associated with the lower-lobe bronchi or arteries within the lung. Paraesophageal nodes. Nodes dorsal to the posterior wall of the trachea and to the right or the left of the midline of the esophagus below the level of the subcarinal region. (Nodes around the descending aorta should also be included.) Right or left pulmonary ligament nodes. Nodes within the right or left pulmonary ligament. Right tracheobronchial nodes. Nodes to the right of the midline of the trachea, from the level of the cephalic border of the azygos vein to the origin of the right-upper-lobe bronchus. Left peribronchial nodes. Nodes to the left of the midline of the trachea, between the carina and the left upper lobe bronchus, medial to the ligamentum arteriosum. Intrapulmonary nodes. Nodes removed in the right or left lung specimen, plus those distal to the main-stem bronchi or secondary carina (includes interlobar, lobar, and segmental nodes). Superior diaphragmatic nodes. Nodes adjacent to the pericardium within 2 cm of the diaphragm.

the anterior portion of the right lung, usually just lateral to the right interlobar pulmonary artery at the level of the bronchus intermedius (24,98). The orientation of the fissure at the level sectioned determines the pattern of the lucent zone, formed by the most peripheral portions of the

Transbronchial needle aspiration Anterior parasternal thoracotomy Percutaneous needle biopsy

adjacent right upper and middle lobes; most often it is triangular, but it may be round or ovoid. The latter two configurations are caused by slight cephalad doming of the fissure, with the right upper lobe completely surrounding the relatively avascular region. With narrow collimation, a portion of the domed minor fissure may be oriented vertical or slightly oblique to the cross-sectional CT plane and may appear as a discrete line or a hazy band on the section. Incomplete fissures are frequently identified (see Fig. 5-58B). In addition to the normal major and minor fissures, accessory fissures may be recognized on CT. Most commonly seen is the previously described azygos fissure (Fig. 5-61), which appears as a laterally convex curved line extending from the right brachiocephalic vein or superior vena cava anteriorly to a posterior position alongside the lateral aspect of the T4 or T5 vertebra. Additional accessory fissures may be formed at the boundaries of bronchopulmonary segments but are seldom visible or recognized on CT scans (94). An inferior accessory fissure (see Fig. 5-58D) may circumscribe the medial or anteromedial basal segment of either lower lobe (11). Also, a left minor fissure, demarcating the lingula from the rest of the left upper lobe, may create a hypovascular area similar in appearance to the right minor fissure (24). The fissures are frequently incomplete, thus allowing the spread of air, infection, or tumor (see Fig. 5-58B). The right major fissure is more often found incomplete than the left. The minor fissure can be incomplete in about 63% of cases (11). The inferior pulmonary ligaments are another normal, frequently recognizable region of the pleura (Fig. 5-62). These ligaments, a double-layer reflection of the parietal pleura, may envelope some bronchial veins and lymph nodes. They anchor the lower lobes to the mediastinum and extend from an apex just below the margins of the inferior pulmonary veins in a caudal and posterior direction. On CT the inferior pulmonary ligaments produce linear shadows arising from the mediastinum and extending laterally into the lung bases (23,55,94,95,230). They may be prominent

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 287

Thorax: Techniques and Normal Anatomy

287

A

B

C

D

Figure 5-57 Nodal stations of calcified mediastinal lymph nodes (see Table 6-2). A: lbv, Level of crossing left brachiocephalic vein; r, right brachiocephalic vein; b, brachiocephalic artery; c, left carotid artery; s, left subclavian artery; arrowhead, normal pretracheal lymph node. B: Level of aortopulmonary window and top of left pulmonary artery; AA, ascending aorta; DA, descending aorta; S, superior vena cava. C: LPA, level of left pulmonary artery; curved arrow, truncus anterior; arrowhead, left upper lobe pulmonary artery. D: l, Level of right interlobar pulmonary artery; aer, azygoesophageal recess. E: Level of cardiac ventricles. IVC, inferior vena cava; e, esophagus; DA, descending aorta; curved arrow, pericardium.

E

5063_Lee_Ch05pp0225-0310 2/17/06 9:49 AM Page 288

288

Chapter 5

A

B

C

D Figure 5-58 Pulmonary fissures. A: The left oblique fissure (arrow) is seen in its upper portion where it is slightly convex posteriorly. B: At midthoracic level, the right oblique fissure shows an incomplete segment (arrow). C: The minor fissure is seen as an area of paucity of vascular markings. D: At a lower level, the oblique fissures are slightly convex anteriorly. Note right inferior accessory fissure (arrowheads) circumscribing the medial basal segment from the rest of the right lower lobe.

in patients with emphysema and can be quite thick in those with pleural thickening, such as secondary to asbestos exposure. On the right side, the inferior pulmonary ligament lies adjacent to the inferior vena cava; on the left side it lies alongside the distal esophagus. Their caudal extensions are variable; they may assume a triangular configuration as they reflect onto the diaphragm or, more rarely, end in a free

falciform edge. The presence of these ligaments explains some of the unusual distributions of fluid or air within the caudal pleural space after lung surgery as well as the configuration of lower lobe atelectasis (163). The inferior phrenic artery and vein can be identified in approximately 30% of helical chest imaging on the right side and up to 42% on the left (265). These vessels

5063_Lee_Ch05pp0225-0310 10/14/05 2:09 PM Page 289

Thorax: Techniques and Normal Anatomy

289

B

A Figure 5-59 A: At level of caudal portion of the heart (H), fat (F) is seen extending into the lower portion of major fissures. B: Same level with lung window settings shows apparent thickening of the major fissure due to extension of fat (F).

manifest as linear opacities extending laterally from the lateral midpoint of the inferior vena cava and from the posterior margin of the left ventricle. In the same region and having a similar appearance is the intersegmental septum of the lower lobe, which is a thin horizontal