Ebook High-Resolution CT of the lung (7/E): Part 1

(BQ) Part 1 book “High-Resolution CT of the lung” has contents: Normal lung anatomy, technical aspects of high-resolution CT, diffuse pulmonary neoplasms and pulmonary lymphoproliferative diseases, pneumoconiosis, occupational, and environmental lung disease, miscellaneous infiltrative lung diseases, cystic lung diseases,… and other contents.

High-Resolution

CT of the Lung

San Francisco, California

Nestor L Müller, MD, PhD

Professor Emeritus of RadiologyDepartment of Radiology, University of British ColumbiaVancouver, British Columbia, Canada

David P Naidich, MD, FACR, FAACP

Professor of Radiology and MedicineNew York University

Langone Medical CenterNew York, New York

F I F T H E D I T I O N

Acquisitions Editor: Ryan Shaw

Product Development Editor: Amy G Dinkel

Production Project Manager: David Orzechowski

Senior Manufacturing Coordinator: Beth Welsh

Marketing Manager: Dan Dressler

Senior Designer: Joan Wendt

Production Service: S4Carlisle Publishing Services

Copyright © 2015 Wolters Kluwer Health

Two Commerce Square

2001 Market Street

Philadelphia, PA 19103 USA

LWW.com

4th edition © 2009 by LIPPINCOTT WILLIAMS & WILKINS, a WOLTERS KLUWER business

All rights reserved This book is protected by copyright No part of this book may be reproduced in any form by any means, including photocopying, or utilized by any information storage and retrieval sys- tem 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 China

Library of Congress Cataloging-in-Publication Data

Webb, W Richard (Wayne Richard), 1945- author.

High-resolution CT of the lung / W Richard Webb, Nestor L Müller, David P Naidich — Fifth edition.

p ; cm.

Includes bibliographical references and index.

ISBN 978-1-4511-7601-8 (alk paper)

I Müller, Nestor Luiz, 1948- , author II Naidich, David P., author III Title.

[DNLM: 1 Lung—radiography 2 Tomography, X-Ray Computed 3 Lung Diseases—pathology

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 infor- mation 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 the publication have Food and Drug Administration (FDA) clearance for limited use in restricted research settings It is the responsibility of the health care providers to ascertain the FDA status of each drug or device planned for use in their clinical practice.

To purchase additional copies of this book, call our customer service department at (800) 638-3030 or fax orders to (301) 223-2320 International customers should call (301) 223-2300.

Visit Lippincott Williams & Wilkins on the Internet: at LWW.com Lippincott Williams & Wilkins customer service representatives are available from 8:30 am to 6 pm, EST.

10 9 8 7 6 5 4 3 2 1

To my father, who encouraged my curiosity and taught

me to figure things out

––WRW

To my wife, Isabela, and my children—Alison, Phillip,

and Noah Müller

––NLM

To Jocelyn, whose constant love and support has always

been my greatest inspiration

––DPN

Salvador, Bahia, Brazil

During the past 25 years, high-resolution CT (HRCT) has

become established as an indispensable tool in the

evalu-ation of patients with diffuse lung disease HRCT is now

commonly used in clinical practice to detect and

char-acterize a variety of lung abnormalities In the

approxi-mately 5 years since our fourth edition was published,

considerable progress has taken place in the

understand-ing of diffuse lung diseases and the recognition of new

entities and their nature, causes, and characteristics

Without doubt, HRCT has played a fundamental role in

contributing to this progress and has become essential to

the diagnosis of a number of diffuse diseases

This fifth edition continues what the three of us,

in-dependently, in conjunction, and with each other’s

en-couragement and support, began some 30 years ago The

photograph of the three of us below was taken by a local

resident at the 1989 Diagnostic Course in Davos, on a walk

we took on the promenade above the Sweitzerhof on the

day of our arrival, when as junior faculty, we were more

than a little anxious about teaching along with such

im-portant and impressive chest radiologists as Fraser, Felson,

Greenspan, Milne, Flowers, Heitzman, and many others

of an inch thick, and, to our knowledge, referenced every known paper on HRCT From our perspective, it was the most important thing we had ever done

That is how things start Maybe that is the best way things should start It was certainly fun and rewarding for each of us And we three have stuck together over the years, out of our combined respect, admiration, friend-ship, and good humor Each one of us believes that we learned more from our collaboration than we taught

In this edition, we have incorporated an update and review of numerous recent advances in the classifica-tion and understanding of diffuse lung diseases and their HRCT features Recent technical modifications in obtain-ing HRCT have also been reviewed, most notably the use

of helical HRCT and dose-reduction techniques We hope the reader will find these changes and updates helpful

As is our wont, we have reorganized our discussions into new sections and chapters, which we feel best presents the most important topics in HRCT diagnosis for reference and learning

A new section has been added at the end of the book

to provide a general review of HRCT, including an trated glossary of HRCT terms and a chapter providing

illus-a compilillus-ation of the common illus-and typicillus-al illus-appeillus-arillus-ances of the most common diffuse lung diseases encountered in clinical practice These sections are intended to provide

an illustrated index to the detailed descriptions of eases found elsewhere in the book

dis-It is with a great deal of pride that we complete our fifth edition of this book, which has occupied so much of our thoughts, efforts, and time over the years This task

is accomplished in the hope that this book will age future generations of thoracic imagers to develop mutually productive relationships with friends and col-leagues, in order to explore important questions in our understanding of the role of imaging in the assessment of thoracic disease

encour-To this end, we acknowledge the contributions of three esteemed colleagues, our former fellows, who have au-thored parts of this book Their efforts have greatly in-spired our own enthusiasm for the considerable task of bringing this edition to fruition

W R ichaRd W ebb N estoR L M üLLeR

d avid P N aidich

At this meeting, we each spoke about the use of HRCT,

which, at the time, was a little-known technique that

was regarded with skepticism by many radiologists We

learned from each other as we spoke, compared slides

in the speaker-ready room, and gained confidence from

our shared opinions At this meeting, we began thinking

about a collaboration that would combine our experience

and thoughts about this new modality and its potential

uses Our first edition of this book was published in late

1991, with a grand total of 159 pages It was a quarter

and prior editions of this book Although they are too numerous to mention here, they are recognized through-out the following pages.

We wish to gratefully acknowledge the many colleagues

who have provided us with insights and inspiration over

the years, and allowed us to use their illustrations for this

Acknowledgments

S E C T I O N I HIgH-REsOLuTION CT TECHNIquEs

1 Technical Aspects of High-Resolution CT 2

S E C T I O N II APPROACH TO HRCT DIAgNOsIs AND

4 HRCT Findings: multiple Nodules and Nodular Opacities 106

6 HRCT Findings: Air-Filled Cystic Lesions 165

S E C T I O N III HIgH-REsOLuTION CT DIAgNOsIs

8 The Idiopathic Interstitial Pneumonias, Part I: usual Interstitial Pneumonia/ Idiopathic Pulmonary Fibrosis and Nonspecific Interstitial Pneumonia 208

9 The Idiopathic Interstitial Pneumonias, Part II: Cryptogenic Organizing Pneumonia, Acute Interstitial Pneumonia, Respiratory Bronchiolitis- Interstitial Lung Disease, Desquamative Interstitial Pneumonia, Lymphoid Interstitial Pneumonia, and Pleuroparenchymal Fibroelastosis 232

14 Hypersensitivity Pneumonitis and Eosinophilic Lung Diseases 376

15 Drug-Induced Lung Diseases and Radiation Pneumonitis 397

16 miscellaneous Infiltrative Lung Diseases 411

17 Infections 429

18 Pulmonary Edema and Acute Respiratory Distress syndrome 481

19 Cystic Lung Diseases 492

20 Emphysema and Chronic Obstructive Pulmonary Disease 517

21 Airways Diseases 552

22 Pulmonary Hypertension and Pulmonary Vascular Disease 622

S E C T I O N IV

23 Illustrated glossary of High-Resolution CT Terms 660

24 Appearances and Characteristics of Common Diseases 678

High-Resolution

CT Techniques and Normal Anatomy

I

S E C T I O N

The first use of the term high-resolution computed

tomography has been attributed to Todo et al (5), who,

in 1982, described the potential use of this technique for assessing lung disease The first reports of HRCT in English date to 1985, including landmark descriptions

of HRCT findings by Nakata et al., Naidich et al., and Zerhouni et al (6–8) Since then, HRCT has become es-tablished as an important diagnostic tool in pulmonary medicine and has significantly contributed to our under-standing of diffuse lung diseases Although many of the HRCT techniques used in these initial studies are still ap-propriate today, the recent development of multidetector helical computed tomography (MDCT) scanners capable

of volumetric high-resolution scanning has significantly changed the manner in which HRCT may be obtained

In this chapter, we review computed tomography (CT) techniques that are appropriate for obtaining HRCT in patients with suspected lung disease, scan protocols rec-ommended in specific clinical settings, the spatial reso-lution and radiation dose associated with HRCT, and common HRCT artifacts

HIGH-RESOLUTION COMPUTED TOMOGRAPHY: FUNDAMENTAL TECHNIQUES

This section reviews the effect of various technical tors on the appearance of HRCT and summarizes our rec-ommendations for obtaining appropriate examinations

fac-HIGH-RESOLUTION COMPUTED TOMOGRAPHY:

FUNDAMENTAL TECHNIQUES 2 TECHNIQUES OF SCAN ACQUISITION: SPACED AXIAL SCANNING VERSUS VOLUMETRIC SCANNING 10 RADIATION DOSE 20

EXPIRATORY HIGH-RESOLUTION COMPUTED TOMOGRAPHY 24

QUANTITATIVE COMPUTED TOMOGRAPHY 30

ASIR adaptive statistical iterative reconstruction

BOS bronchiolitis obliterans syndrome

COPD chronic obstructive pulmonary disease

CTDI CT dose index

DLP dose length product

ECG electrocardiographic

FBP filtered back projection

FOV field of view

MinIP minimum-intensity projection

MBIR model-based iterative reconstruction

MDCT multidetector helical computed tomography

MD-HRCT multidetector helical HRCT

NSIP nonspecific interstitial pneumonia

ROI region of interest

3D three-dimensional

2D two-dimensional

Abbreviations Used in This Chapter

High-resolution computed tomography (HRCT) is capable

of imaging the lung with excellent spatial resolution,

pro-viding anatomical detail similar to that available from gross

pathologic specimens and paper-mounted lung slices (1–4)

HRCT can readily demonstrate the normal and abnormal

C H A P T E R 1 Technical Aspects of High-Resolution CT 3

Although each author performs HRCT in a different

manner, we generally agree as to what fundamental

tech-niques constitute a “high-resolution” CT study Quite

simply, these include (a) the use of thin-collimation axial

scans or thin-section reconstruction of volumetric data

obtained using MDCT and narrow detector width (0.5–

1.25 mm) and (b) image reconstruction with a high spatial

frequency (sharp or high-resolution) algorithm Sufficient

radiation (in milliampere seconds [mAs] or effective mAs

[mAs/pitch for helical scans]) (9) must be used to keep

image noise at a level low enough to allow accurate image

interpretation, while keeping patient exposure at

appro-priate levels; keep in mind that dose reduction techniques

can be used while obtaining diagnostic scans (Table 1-1)

(1–4,10–12) Targeted image reconstruction may be used

to reduce pixel size, but is not necessary for clinical

diag-nosis in most settings (Table 1-1) (1–4,10–12)

Slice Thickness

The use of thin sections (0.5–1.5 mm) is essential if spatial

resolution and lung detail are to be optimized (4,6,8,10)

(Table 1-1) Generally, 1-mm-thick slices are adequate for

diagnosis; a clear-cut advantage for thinner slices has not

been shown (13) With slices thicker than 1 to 1.5 mm,

volume averaging within the plane of scan significantly

reduces the ability of CT to resolve small structures The use of 2.5- to 5-mm slice thickness should not be consid-ered adequate for HRCT

In an early study, Murata et al (12) compared the ability of axial HRCT performed with 1.5- and 3-mm collimation to allow the identification of small vessels, bronchi, interlobular septa, and some pathologic findings With 1.5-mm collimation, greater contrast was evident between vessels and surrounding lung parenchyma, more branches of small vessels were seen, and small bronchi were more often recognizable than with 3-mm collima-tion (12) Also, slight increases in lung attenuation (as may be seen in early interstitial lung disease), or decreases

in attenuation (as in emphysema), were better resolved with 1.5-mm collimation However, the authors con-cluded that certain pathologic findings, such as thickened interlobular septa, were similarly visible on images with 1.5- and 3-mm collimation (12)

There are several differences in how lung structures are visualized on scans performed with thin (e.g., 1-mm) and thick (e.g., 5-mm) sections With thin slices, it is more dif-ficult to follow the courses of vessels and bronchi than it

is with thick slices With thick slices, for example, vessels that lie in the plane of scan look like vessels (i.e., they appear cylindrical or branching) and can be clearly identi-fied as such With thin slices, vessels can appear round or

Recommended

Slice thickness: thinnest available (0.5–1.5 mm)

Reconstruction algorithm: high spatial frequency or “sharp” algorithm

kV(p) 120; 100 or 80 for small or pediatric patients

mA less than 250; mAs (effective) of 100 or less

Scan (rotation) time: as short as possible (e.g., 0.3–0.5 s)

Pitch (MD-HRCT): 1-1.5

Inspiratory level: full inspiration

Position: supine; prone scans routinely in patients with suspected interstitial lung disease; in patients with minimal or unknown chest film

abnormalities, or monitor supine scans for dependent density

Acquisition: spaced axial imaging or MD-HRCT

Expiratory imaging: postexpiratory scans at three or more levels in patients with obstructive disease

Reconstruction: transaxial; entire thorax

Windows: at least one consistent lung window setting is necessary Window mean/width values of 600–700 HU/1,000–1,500 HU are appropriate Good combinations are 700/1,000 HU or 600/1,500 HU Soft-tissue windows of approximately 50/350 HU should also be used for the

mediastinum, hila, and pleura.

Image display: workstation (optimal) or photography of lung windows 12 on 1

Optional

Reduced mAs: low-dose axial HRCT or MD-HRCT best for follow-up studies

Acquisition: ECG gating or segmented reconstruction to reduce motion artifacts

Expiratory imaging: dynamic, volumetric, or spirometrically triggered expiratory scans

Contrast injection: patients with suspected vascular disease

Reconstruction: targeted (15- to 25-cm FOV; 2D or 3D reconstruction; MIP or MinIP reconstructions)

Windows: windows may need to be customized; a low window mean (800–900 HU) is optimal for diagnosing emphysema For viewing the

mediastinum, 50/350 HU is recommended For viewing pleuroparenchymal disease, 600/2,000 HU is recommended

TABLE 1-1 Summary of HRCT Techniques

FIGURE 1-1 Effect of slice thickness on resolution A: Helical CT with 5-mm slice thickness, reconstructed with the standard algorithm in a normal subject A number of

branching pulmonary vessels are visible (arrows) B: Helical CT at the same level with 1.25-mm slice thickness reconstructed with the same scan data and algorithm Pulmonary

arteries seen as branching or cylindrical on the thicker scan appear “nodular” on the scan with 1.25-mm slice thickness (arrows) The resolution is clearly improved with thin slices.

oval (i.e., nodular) because only short segments may lie

in the plane of scan (Fig 1-1) With experience, this

dif-ficulty is easily avoided

Also, with thin slices, the diameter of a vessel that lies

in or near the plane of scan can appear larger than it does

with thicker slices because less volume averaging occurs

between the rounded edge of the vessel and the adjacent

air-filled lung; thin scans more accurately reflect vessel

diameter in this setting, analogous to the better

estima-tion of the diameter of a lung nodule that is possible with

thin slices Furthermore, with thin slices, bronchi that are

oriented obliquely relative to the scan plane are much

better defined than they are with thicker slices, and their

wall thicknesses and luminal diameters are more

accu-rately assessed (14) The diameters of vessels or bronchi

that lie perpendicular to the scan plane appear the same

with both thin and thick collimation

Reconstruction Algorithm

The inherent or maximum spatial resolution of a CT

scan-ner is determined by the geometry of the data-collecting

system and the frequency at which scan data are sampled

during the scan sequence (10) The spatial resolution of

the image produced is less than the inherent resolution of

the scan system, depending on whether axial or

volumet-ric (helical) imaging is used, the reconstruction algorithm,

the matrix size, and the field of view (FOV), all of which

in turn determine pixel size In HRCT, these parameters

are optimized to increase the spatial resolution of the

image

With body CT, scan data are usually reconstructed

with a relatively low spatial frequency algorithm (e.g.,

“standard” or “soft-tissue” algorithms) that smoothes the

image, reduces visible image noise, and improves the

con-trast resolution to some degree (11,15) Low spatial

fre-quency simply means that the frefre-quency of information

recorded in the final image is relatively low; it is the same

as saying that the algorithm is low resolution rather than high resolution

Reconstruction of images using a sharp, high spatial frequency, or high-resolution algorithm reduces image smoothing and increases spatial resolution, making struc-tures appear sharper (Figs 1-2 to 1-4) (6,10,12,16) Us-ing a high-resolution algorithm is a critical element in performing HRCT (Table 1-1) (11,15) In one study of HRCT techniques (10), the use of a high spatial frequency algorithm to reconstruct scan data resulted in a quantita-tive improvement in spatial resolution when compared to

a standard algorithm (Fig 1-3); in this study, subjective image quality was also rated more highly with the high spatial frequency algorithm In another study of HRCT (12), small vessels and bronchi were better seen when im-ages were reconstructed with a high-resolution algorithm than when the standard algorithm was used The use of a sharp algorithm has also been recommended to improve spatial resolution for routine chest CT reconstructed with thicker slices (17)

Kilovolts (Peak), Milliamperes, and Scan Time

Using a sharp or high-resolution reconstruction rithm, in addition to increasing image detail, increases the visibility of noise in the CT image (11,15) This noise usually appears as a graininess, mottle, or streaks that can be distracting and may obscure anatomical detail (Fig 1-4) (10) Because much of this noise is quantum related, it is inversely proportional to the number of photons absorbed (precisely, it is inversely proportional

algo-to the square root of the product of mA and scan time) (16) Consequently, it increases with decreasing mAs or kilovolt peak (kV(p)) and decreases with increased mAs

or kV(p) (Fig 1-5) (10,16) For example, in one study

C H A P T E R 1 Technical Aspects of High-Resolution CT 5

been reconstructed using a high-resolution (sharp) algorithm (A) and a smooth (standard) algorithm (B) Lung structures, reticular opacities, and traction bronchiectasis are

much more sharply defined with the high-resolution algorithm.

standard algorithm Numbers indicate the resolution in line pairs per centimeter The resolution with this technique is 6 line pairs per centimeter B: When the same scan

is reconstructed using the high-resolution (i.e., bone) algorithm, spatial resolution improves Also, in contrast to the scan reconstructed using the standard algorithm, 7.5

line pairs are easily resolved (arrow), and edges are considerably sharper (From Mayo JR, Webb WR, Gould R, et al High-resolution CT of the lungs: an optimal approach

Radiology 1987;163:507, with permission.)

using an early-generation scanner (10), a measure of

image noise was reduced by approximately 30% when

kV(p)/mAs were increased from 120/200 to 140/340

(Fig 1-5), and the scans with increased kV(p) and mAs

settings were rated as being of better quality in 80% of

cases (Fig. 1-6) (10)

Although increasing mAs or kV(p) above routine

val-ues can reduce image noise, it is not necessary for

obtain-ing adequate HRCT images, and maintenance of patient

radiation dose at a reasonable level is considered to be

more important (16) With current scanners and

recon-struction algorithms, diagnostic scans can be obtained

using mAs and kV(p) techniques considered routine for

chest CT Scan techniques with a kV(p) of 120 are

gener-ally used, although a reduced kV(p) of 100 or 80 may be

used in small or pediatric patients (i.e., less than 80 or

60 kg) (13)

Using mAs (or effective mAs) values of 100 or less has proven satisfactory for obtaining HRCT in most patients with current-generation scanners (13,18) Increased pa-tient size and increased chest wall thickness are associ-ated with increased image noise; this may be reduced with increased mA (Fig 1-7) (10) Reducing mA to 40 (i.e., low-dose CT) may be used to reduce image dose, but this should generally be reserved for small or pediatric pa-tients Image noise may be excessive with low mA settings

in large patients (Fig 1-8)

Specific mA, kV(p), pitch (with helical scanning), and gantry rotation times most appropriate for HRCT vary with different scanners When obtaining helical HRCT,

FIGURE 1-4 Effect of reconstruction algorithm on resolution and image noise A 1.25-mm MD-HRCT has been reconstructed with high-resolution (A) and standard (B) algorithms A: The image reconstructed with the high-resolution algorithm is sharper and shows more detail, but streak artifacts due to aliasing and noise are more

apparent B: Resolution is diminished with this algorithm The image appears smoother with this algorithm, and noise is less apparent.

HRCT image noise (SD of HU measurements) in an anthropomorphic CT phantom

as related to the reconstruction algorithm and scan technique Noise increases when

the bone (high-resolution) algorithm is used instead of the standard algorithm With

the bone algorithm, noise decreases approximately 30% with increased kV(p) and mA

settings (From Mayo JR, Webb WR, Gould R, et al High-resolution CT of the lungs: an

optimal approach Radiology 1987;163:507, with permission.)

the use of dynamic, modulated, or adaptive mA that

var-ies with body thickness should generally be used to keep

radiation dose low, without sacrificing image quality (19)

In large patients, a reasonable maximum mA should be

set when using this technique, to avoid inappropriately

high exposures

Because of artifacts related to patient motion,

breath-ing, and cardiac pulsation, it is desirable to minimize scan

or gantry rotation time A scan time or gantry rotation

time of 0.5 s or less is optimal for HRCT and, if available,

is recommended (Table 1-1) Most current scanners have

gantry rotation times of 300 to 500 ms

Field of View and Targeted Reconstruction

Scanning should be performed using the smallest FOV that

will encompass the patient (e.g., 35 cm), as this reduces

pixel size Retrospectively targeting image reconstruction

to a single lung instead of the entire thorax significantly

obtained with a tube current of 100 mA (A) and 400 mA (B) in a patient with atypical

mycobacterial infection There is a relative increase in noise in A, which is evident

both in the soft tissues and lung Note, however, that the lower-dose scan (A) is still

of diagnostic quality.

A

B

C H A P T E R 1 Technical Aspects of High-Resolution CT 7

Images through the upper (A), mid- (B), and lower (C) lungs are shown from a

normal HRCT obtained at 1-cm intervals in the supine position in inspiration using

a fixed tube current (40 mA) Dynamic expiratory images were also obtained at three selected levels The estimated effective dose for this examination was 0.2 mSv However, image noise is excessive, and subtle abnormalities may be difficult

to detect.

A

B

C

measured using an anthropomorphic chest phantom, with simulated thick and thin

chest walls Noise significantly increases with the thick chest wall (From Mayo JR,

Webb WR, Gould R, et al High-resolution CT of the lungs: an optimal approach

Radiology 1987;163:507, with permission.)

Thick chest wall

reduces the FOV and image pixel size, and thus increases

spatial resolution (Figs 1-9 and 1-10) (10,20,21) For

example, with a 40-cm reconstruction circle (FOV) and

a 512 × 512 matrix, pixel size measures 0.78 mm With

targeted image reconstruction using a 25-cm FOV, pixel

size is reduced to 0.49 mm, and the spatial resolution is

correspondingly increased (Fig 1-9) Using a 15-cm FOV

further reduces pixel size to 0.29 mm, but this FOV is

usually insufficient to view an entire lung and is not often

used clinically It should be recognized, however, that the

improvement in resolution obtainable by targeting is

lim-ited by the intrinsic resolution of the detectors used

The use of targeted reconstruction is often a matter of

personal preference In clinical practice, the use of image

targeting is uncommon because it requires additional

re-construction time, the raw scan data must be saved until

targeting is performed, and display of the individual lung

images is somewhat cumbersome With a nontargeted

re-construction, the ability to see both lungs on the same

image allows a quick comparison of one lung to the other;

this can be quite helpful in diagnosis and is preferred to the

marginal increase in resolution achieved with targeting

Inspiratory Level

Routine HRCT is obtained during suspended full

in-spiration, which (a) optimizes contrast between normal

structures, various abnormalities, and normal aerated

lung parenchyma; and (b) reduces transient atelectasis, a

finding that may mimic or obscure significant

abnormali-ties Selected scans obtained during or after forced

expi-ration may also be valuable in diagnosing patients with

obstructive lung disease or airway abnormalities The use

of expiratory HRCT is discussed later in this chapter, and

in Chapters 2 and 7

FIGURE 1-9 Effect of targeted reconstruction on resolution A: HRCT image in a patient with end-stage sarcoidosis obtained with a 38-cm FOV and 1.5-mm

collimation, and reconstructed using a high-resolution algorithm and a 38-cm reconstruction circle B: The same CT scan has been reconstructed using a targeted FOV

(15 cm), reducing image pixel diameter Image sharpness is improved compared to A.

using a targeted FOV of 25 cm The resolution with this technique is 7.5 line pairs (arrow) B: The same scan viewed without targeting shows the effects of larger pixel size

Only 6 line pairs can be resolved (arrow), and the margins of the lines appear jagged or wavy (From Mayo JR, Webb WR, Gould R, et al High-resolution CT of the lungs: an optimal approach Radiology 1987;163:507, with permission.)

Patient Position and the Use

of Prone Scanning

Scans obtained with the patient supine are adequate for

diagnosis in most instances However, scans obtained

with the patient positioned prone are sometimes

neces-sary for diagnosing subtle lung abnormalities Atelectasis

is commonly seen in the dependent lung (i.e., posterior

lung on supine scans) in both normal and abnormal

subjects, resulting in a so-called dependent density or

subpleural line (Fig 1-11) (22,23) These normal ings can closely mimic the appearance of early lung fibro-sis, and they can be impossible to distinguish from true pathology on supine scans alone However, if scans are obtained in both supine and prone positions, dependent density can be easily differentiated from true pathol-ogy Normal dependent density disappears in the prone position (Fig 1-11); a true abnormality remains visible regardless of whether it is dependent or nondependent (Figs 1-12 and 1-13)

C H A P T E R 1 Technical Aspects of High-Resolution CT 9

in a subpleural region anteriorly B: On a prone image, the posterior lung is unchanged in appearance, indicative of lung disease.

appears normal Note that some dependent opacity is now visible in the anterior lung.

Dependent density results in a diagnostic dilemma

only in patients who have normal lungs or subtle lung

abnormalities In patients with obvious abnormalities,

such as honeycombing, or in patients with diffuse lung

disease, dependent density is not usually a diagnostic

problem Thus, if the patient being studied has evidence

of moderate- to-severe lung disease on plain radiographs,

prone scans are not likely to be needed However, if the

patient is suspected of having an interstitial

abnormal-ity and the plain radiograph is normal or near normal,

or the results of chest radiographs are unknown, prone

scans may prove helpful In addition, even in patients

with obvious lung disease on supine scans, prone scans

may prove useful in identifying specific important

diag-nostic findings (i.e., subtle posterior lung honeycombing),

not clearly seen on the supine images

Volpe et al (24) assessed the usefulness of prone scans

in patients who had chest radiographs read as normal,

possibly abnormal, or definitely abnormal Overall, prone

scans were considered helpful in 17 of 100 consecutive

patients having HRCT (24) Prone HRCT scans were helpful in confirming or ruling out posterior lung abnor-malities in 10 of 36 (28%) patients who had normal find-ings on chest radiographs, 5 of 18 (28%) patients who had possibly abnormal findings on chest radiographs, and only 2 of 46 (4%) patients who had definitely abnormal findings on chest radiographs The proportion of patients who benefited from prone scans was significantly lower among the patients with abnormal findings on chest

radiographs than among the patients with normal (p = 0.008) or possibly abnormal (p = 0.02) findings The two

patients who had abnormal findings on radiographs and

in whom CT scans obtained with the patient prone were helpful had minimal radiographic abnormalities

Some investigators (21,25) obtain HRCT in the prone position only when dependent lung collapse

is problematic (26); however, this approach requires that the scans be closely monitored or that the patient

be called back for additional scans Others use prone scanning in specific clinical settings, for example, when

FIGURE 1-13 Persistent posterior lung ground opacity on prone scans in a patient with scleroderma and NSIP A: Supine scan shows ill-defined opacity

in the posterior lungs B: On a prone image, the posterior subpleural lung opacity is unchanged in appearance, and the presence of true lung disease can be

diagnosed.

asbestosis or early lung fibrosis is suspected, whereas

still others obtain prone scans routinely (22,27) In

patients who are suspected of having emphysema,

airways disease such as bronchiectasis, or another

obstructive lung disease, dependent atelectasis is not

usually a diagnostic problem, and prone scans are not

usually needed

Spaced axial prone scans, prone scans clustered near

the lung bases, or volumetric helical imaging in the

prone position may all be used Some protocols call for

prone volumetric imaging only (i.e., no supine scans are

obtained) (28); this would be most useful in a patient

suspected of having a disease with a posterior lung

pre-dominance, such as asbestosis or idiopathic pulmonary

fibrosis

TECHNIQUES OF SCAN ACQUISITION:

SPACED AXIAL SCANNING VERSUS

VOLUMETRIC SCANNING

Before the introduction of MDCT scanners, HRCT was

performed by obtaining individual scans at spaced

inter-vals This technique remains in use today However, the

development of MDCT scanners, capable of rapidly

im-aging the thorax using an isotropic technique, has greatly

expanded the ways in which a HRCT study may be

obtained (13)

Spaced Axial Scans

HRCT may be performed with individual axial scans

be-ing obtained at spaced intervals, usually 1 to 2 cm,

with-out table motion (Figs 1-14 and 1-15) In this manner,

HRCT is intended to “sample” lung anatomy, with the

assumptions being that (a) a diffuse lung disease will

be visible in at least one of the levels sampled and (b) the

findings seen at the levels scanned will be representative

of what is present throughout the lung These

assump-tions have proven valid during more than 20 years of

experience with HRCT (29)

(A) and 1.25-mm MD-HRCT (B) in a patient with scleroderma and fibrotic NSIP

Two prone HRCT images at the same level are shown in a patient with related NSIP While of similar diagnostic quality, the axial HRCT (A) has slightly better

scleroderma-resolution and the structures and abnormalities appear sharper than on the helical HRCT (B).

A

B

When spaced axial scanning is chosen for HRCT,

we consider scans obtained at 1-cm intervals, from the lung apices to bases, to be the most appropriate routine

C H A P T E R 1 Technical Aspects of High-Resolution CT 11

1.25-mm MD-HRCT (B and C) in a patient with mixed connective tissue disease

and NSIP A: Axial 1.25-mm HRCT shows irregular reticulation, ground-glass

opacity, and traction bronchiectasis with lower-lobe predominance Subpleural sparing is present These findings are typical of NSIP 2D and 3D reconstructed images from the MD-HRCT are also shown in Fig 1-16 B and C: Comparable

levels from the MD-HRCT show identical findings There is no significant difference

in diagnostic value of the axial and MD-HRCT images, although the MD-HRCT images appear slightly smoother.

A

C

B

scanning protocol, allowing an adequate sampling of the

lung and lung disease regardless of its distribution

In early reports, HRCT scanning was sometimes

per-formed with scans at 2-, 3-, and even 4-cm intervals

(3,26); at three preselected levels (25); or at one or two

levels through the lower lungs (21) Although such wide

spacing may be sufficient for assessing some patients

and some lung diseases, in many cases, these protocols

would prove inadequate for initial diagnosis It should

be pointed out, however, that in patients with a known

disease, a limited number of HRCT images may be

suf-ficient to assess disease extent For example, in one study

(30), the ability of HRCT obtained at three selected levels

(limited HRCT) to show features of idiopathic

pulmo-nary fibrosis was compared to that of HRCT obtained

at 10-mm increments (complete HRCT) HRCT fibrosis

scores strongly correlated with pathology fibrosis scores

for both the complete (r = 0.53, p = 0.0001) and

lim-ited (r = 0.50, p = 0.0001) HRCT examinations HRCT

ground-glass opacity scores also correlated with the

his-tologic inflammatory scores on the complete (r = 0.27,

p  = 0.03) and limited (r = 0.26, p = 0.03) HRCT

ex-aminations Similarly, in evaluating patients with

asbes-tos exposure, several investigators have suggested that a

limited number of scans should be sufficient for the

diag-nosis of asbestosis (22,27,31–34) Obtaining four or five

scans near the lung bases has proved to have good

sen-sitivity in patients with suspected asbestosis (35)

Thick-slice CT, combined with a few HRCT images, has also

been applied to patients with suspected diffuse lung ease and has been shown to be clinically efficacious (35); HRCT scans obtained at the levels of the aortic arch, ca-rina, and 2 cm above the right hemidiaphragm allow the assessment of the lung regions in which lung biopsies are most frequently performed (11)

dis-In patients who are likely to require prone images, prone scans can be added to the routine supine sequence obtained with 1-cm scan spacing; a reasonable protocol would include additional prone scans at 2-cm intervals Although axial imaging is a low-dose technique, if fur-ther radiation dose reduction is a concern, scans could be obtained at 2-cm intervals from lung apices to bases, in both supine and prone positions Because the prone and supine images will be slightly different, even if an attempt

is made to obtain the scans at exactly the same levels, the number of different levels scanned will be equivalent to a supine position scan protocol using 1-cm spacing

Another dose reduction technique used with axial aging is to customize the number or location of scans, depending on the patient’s suspected disease, clinical findings, or the location of plain radiographic abnor-malities For example, if the lung disease being studied predominates in a certain region of lung, as determined

im-by chest radiographs, conventional CT (21), or other imaging studies, it makes sense that more scans should

be obtained in the most abnormal area In patients with suspected asbestosis, it has been recommended that more scans be performed near the diaphragm than in the upper

lobes because of the typical basal distribution of this

dis-ease, even if the chest radiograph does not suggest an

ab-normality in this region (22,27)

Some support for this approach has been lent by a

pa-per (36) describing theoretical methods useful in selecting

the appropriate number of HRCT images for estimating

any quantitative parameter of lung disease A marked

re-duction in the number of images necessary for

quantifica-tion of a desired parameter can be achieved by using a

stratified sampling technique based on prior knowledge

of the disease distribution

Spaced axial HRCT scans may be obtained in

combi-nation with a volumetric helical CT study, in which the

entire thorax is imaged (11,21,25), although this is rarely

needed in current practice If both volumetric imaging

and HRCT are needed for diagnosis, it is usually adequate

to reconstruct the volumetric scan with thin slices and a

sharp algorithm

However, Leswick et al (37) compared the patient

ra-diation dose from a combination of spaced axial HRCT

and volumetric helical MDCT to that from a volumetric

helical HRCT having a noise level similar to that of the

axial scans The authors found that the volumetric helical

HRCT had a radiation dose 32% higher than that in the

combined study (37)

Volumetric High-Resolution

Computed Tomography

The use of MDCT scanners capable of rapid scanning and

thin-slice acquisition has revolutionized HRCT technique

Volumetric HRCT using thin detectors (0.5–0.625 mm)

has become the routine in many institutions

In an early attempt at volumetric imaging (38), four

contiguous HRCT scans were obtained without using

helical technique at each of three locations (the aortic

arch, carina, and 2 cm above the right hemidiaphragm)

in 50 consecutive patients with interstitial lung disease or

bronchiectasis At each level, the diagnostic information

obtainable from the set of four scans was compared to

that obtainable from the first scan in the set of four When

the full set of four scans was considered, more findings of

disease were identified The sensitivity of the first scan as

compared to the set of four was 84% for the detection

of bronchiectasis, 97% for ground-glass opacity, 88%

for honeycombing, 88% for septal thickening, and 86%

for nodular opacities (38) However, it is more likely that

the improvement in sensitivity found using the set of four

scans reflects the number of scans viewed rather than the

fact that they were obtained in contiguity

Although volumetric HRCT images appear slightly

smoother than axial HRCT (Figs 1-14 and 1-15), the

technique has several advantages It allows (a) complete

imaging of the lungs and thorax, (b) viewing of contiguous

slices for the purpose of better defining lung

abnormali-ties, (c) reconstruction of scan data in any plane or using

maximum-intensity projections (MIPs) or

minimum-intensity projections (MinIPs), (d) precise level-by-level

comparison of studies obtained at different times for uation of disease progression or improvement, and (e) the diagnosis of additional thoracic abnormalities (Figs 1-16

eval-to 1-24) On the other hand, the use of volumetric tidetector helical HRCT (MD-HRCT) results in a greater radiation dose than does spaced axial imaging

mul-It is not unusual for only one or two slices from a metric HRCT study to provide the key observations nec-essary for diagnosis (39–44) For example, in a patient with suspected idiopathic pulmonary fibrosis, the pres-ence of honeycombing, which is necessary for a definite diagnosis, may be visible only on a few images This find-ing could be missed on spaced axial images

volu-MDCT scanners make use of multiple adjacent tor rows that acquire scan data simultaneously and may

detec-be used independently or in combination to generate ages of different thickness (45) Current MDCT scan-ners are capable of imaging the entire thorax within a few seconds, with the volumetric reconstruction of thin, high-resolution slices For example, using a 64-detector scanner, data may be simultaneously acquired from sixty-four 0.625-mm-thick detector arrays, a pitch of 1 to 1.5, and a gantry rotation time of 0.5 s or less The volumetric data resulting from this mode of scanning allow isotropic imaging and HRCT assessment of lung morphology in

im-a continuous fim-ashion from lung im-apex to bim-ase, the ing of the scan volume in nontransaxial planes or with three-dimensional (3D) reconstruction (Figs 1-16A–C to 1-18), and the production of MIP and MinIP images at any desired level or in alternate planes (Figs 1-16D–F and 1-19 to 1-24) This technique also allows a volumetric

view-CT examination of the thorax to be easily combined with HRCT

Even with a rapid scanner, dyspneic patients with fuse lung disease may not be able to hold their breath for the duration of a volumetric study In such patients,

dif-if optimal resolution is desired, the scan protocol may be modified according to the distribution of the disease sus-pected For diseases likely to have a basal predominance, such as idiopathic pulmonary fibrosis, scanning should begin near the diaphragm and proceed cephalad In this manner, the more important basal lung will be imaged

at the beginning of the scan sequence, and if the patient begins to breathe during scanning, only images through the less important upper lobes will be degraded by respi-ratory motion For the same reason, in a patient suspected

of having a disease with an upper-lobe predominance (e.g., sarcoidosis), it is appropriate to begin scanning in the lung apices Because lung movement with respira-tion is greatest at the lung bases, an alternative approach would be to scan from the bases to apices in all patients

If the patient breathes during the scan, the upper lobes would be less affected

The helical acquisition of HRCT data results in some broadening of the scan profile as compared to detector width Using a low value of pitch (e.g., 1) is recommended

to minimize this effect (29) However, the effective slice thickness obtained using MD-HRCT is clearly sufficient for

connective tissue disease and NSIP This is the same patient as shown in Fig 1-15

A: 2D coronal reconstruction shows findings identical to those in Fig 1-15 The

basal distribution of the abnormalities is well shown, but the same information would be available from review of the transaxial images B: 2D sagittal

reconstruction clearly shows the posterior and lower-lobe predominance of the abnormalities, lower-lobe volume loss as evidenced by posterior displacement

of the major fissure (white arrows), and subpleural sparing (black arrows)

C: 3D surface-display reconstruction with a perspective from below the lung bases

shows the distribution of basal-predominant lung disease, but otherwise is of little diagnostic value D: Transaxial MIP image at the same level as shown in Fig 1-15C

MIP imaging in this patient obscures detail and is of little diagnostic value E and F: 3D MinIP coronal (E) and sagittal (F) reconstructions show the airways and

lower-lobe traction bronchiectasis to best advantage Ground-glass opacity is less apparent in the lung bases than on the routine images.

HRCT diagnosis when thin detectors are used (Figs. 1-14

and 1-15) With 0.625-mm detector width and a pitch

of 1, the effective slice thickness is 1 mm or less

Practi-cally speaking, in most situations, using thin detectors and

standard pitch is adequate for diagnosis

Depending on the technique used and how data from

the various detector rows are combined, images of

differ-ent thickness may be produced retrospectively from the

same study Using the protocol described previously, in

addition to viewing images generated from data acquired

by the individual detectors, data from the detector rows

may be combined to produce images representing thicker

slices (i.e., 2.5 or 5 mm) Thus, this technique enables

HRCT and “routine” or thick-section chest imaging to be

combined as a single examination, blurring the

distinc-tion between these studies

Combining a volumetric chest CT examination with

HRCT by using MDCT may be of value in patients

be-ing studied primarily for diffuse lung disease, for which

HRCT would be the examination of choice, and in

pa-tients being evaluated for a disease or abnormality

usu-ally studied using a thicker-slice helical CT For example,

in patients with hemoptysis, both thin and thick image

reconstruction may be of value in demonstrating both

small or large airways disease and vascular abnormalities (46) Another advantage of MDCT would be in patients requiring CT for the diagnosis of thoracic disease such a lung carcinoma In such patients, scan data may be recon-structed with a thickness appropriate for the detection of lung nodules and bronchial abnormalities and for assess-ment of mediastinal and hilar lymph nodes At the same time, and without additional scanning, high-resolution images could be reconstructed for the purpose of delin-eating nodule morphology and attenuation, or for the diagnosis of associated lymphangitic spread of carcinoma.Similarly, in patients with suspected pulmonary vascu-lar disease, HRCT with contrast enhancement may be ob-tained using MDCT, allowing the detailed assessment of both vasculature and pulmonary parenchyma (Figs. 1-18 and 1-19) (46,47) In patients having helical CT for the diagnosis of acute or chronic pulmonary embolism or pulmonary hypertension, scan data can be reconstructed using different algorithms to look for vascular abnormali-ties and lung disease that could be associated with similar symptoms

Although it is clear that MDCT scanners produce nostic helical HRCT examinations, the results of studies comparing MD-HRCT to spaced axial HRCT have been mixed, and neither imaging method appears to offer a clear advantage For example, Sumikawa et al (48) found that the quality of MD-HRCT images was equivalent to that of axial HRCT in 11 autopsy lungs; visualization of abnor-mal structures and diagnostic efficacy with MD-HRCT (0.75-mm collimation, pitch of 1) was equal to that of axial scans with 0.75-mm collimation Also, Schoepf et al (49) compared MD-HRCT using a 1.25-mm detector and

diag-a pitch of 1.5 with spdiag-aced diag-axidiag-al imdiag-ages (1-mm slices) in

two groups of patients No significant difference (p = 0.986) was found between multislice and single-slice axial HRCT sections in an overall score of image quality, spa-tial resolution, subjective signal-to-noise ratio, diagnostic value, depiction of bronchi and parenchyma, and motion and streak artifacts (49) In contrast, Honda et al (50) compared the image quality and diagnostic efficacy of MD-HRCT (1.25-mm slices) to axial HRCT (1-mm slices)

in imaging cadaveric lungs The image quality of  axial HRCT was considered superior to that of MD-HRCT ob-tained with 1.25-mm detectors and a pitch of 0.75 or 1.5, and less image noise was present with axial HRCT How-ever, the diagnostic efficacy of MD-HRCT with a pitch of 0.75 was equal to that of axial HRCT (50)

Kelly et al (51) found that MD-HRCT may be ated with significantly greater motion artifact compared with axial HRCT obtained in the same patient However, MD-HRCT scans were obtained using 4- or 8-detector scanners, and scan time was undoubtedly longer than that with current MDCT scanners On the other hand, Studler

associ-et al (52) found that motion artifacts were significantly more common on axial HRCT scans (1-mm collimation) than on MD-HRCT (1.5-mm detectors, pitch 1.25) im-

ages (p < 0.001) However, the authors believed that the

assessment of ground-glass opacity was superior on axial F

pneumonia This represents the same patient as shown in Fig 1-2 A: Transaxial

MD-HRCT with 1.25-mm slice thickness shows reticulation, traction bronchiectasis, and ground-glass opacity with a posterior and subpleural distribution The upper lobes were less abnormal B: Sagittal reconstruction (0.7 mm thick) shows these

abnormalities to predominate in the posterior and basal subpleural lung (arrows)

C: Coronal reconstruction (0.7 mm thick) through the posterior lung shows a

similar distribution.

HRCT The effective radiation doses were 3.8 millisievert

(mSv) for MDCT and 0.9 mSv for axial HRCT When

considering the relative value of spaced axial HRCT and

HRCT, the greater radiation dose involved in

MD-HRCT should be kept in mind

Benaoud et al (53) compared 1-mm-thick images

re-constructed contiguously through the chest with images

spaced at 1-cm intervals in the evaluation of chronic

bronchopulmonary diseases The spaced reconstructions

would have resulted in a 79% radiation reduction, and

there was almost perfect agreement (kappa = 0.83–1) for

both the detection and distribution and findings between

the volumetric and spaced reconstructions (53) In the

screening of patients with scleroderma, Winklehner et al

(54) showed equal sensitivity for interstitial lung disease

comparing volumetric MD-HRCT and spaced 1-mm

sec-tions reconstructed at 1-cm intervals Certain HRCT

find-ings, however, that have a heterogenous distribution may

be better detected and quantified using volumetric HRCT

For instance, in the assessment of bronchiolitis obliterans

syndrome (BOS) in lung transplant recipients, Dodd et al

(55) showed that volumetric MDCT correlated with the

stage of BOS, whereas spaced HRCT images did not

Despite these limitations, volumetric HRCT is ing standard in many patients, for many indications, and

becom-in many becom-institutions At least partially, this reflects recent advances in radiation dose reduction with volumetric CT Often, only the supine images will be obtained with volu-metric imaging technique

Reconstruction Techniques with Volumetric High-Resolution Computed Tomography

Sagittal and Coronal Reformations

MD-HRCT produces isotropic scans, allowing ous 3D visualization of the lung parenchyma and the ca-pacity to create high-quality two-dimensional (2D) and 3D reformatted images (Figs 1-16 to 1-20) (20) Honda

contigu-et al compared the quality of coronal multiplanar structions obtained from an MDCT data set (0.5-mm col-limation, 0.5-mm reconstruction interval) with the quality

recon-of direct coronal MD-HRCT (0.5-mm collimation) scans

in 10 normal autopsy lung specimens Image quality was considered equal (56) It is clear that MD-HRCT

FIGURE 1-18 Contrast-enhanced MD-HRCT in an AIDS patient with pulmonary hypertension, obtained with 1.25-mm slice thickness The differential diagnosis included chronic pulmonary embolism, vasculitis, and lung disease Transaxial (A and B) and sagittal reconstructed (C) HRCTs obtained during a single breath hold show

normal findings D: Transaxial image shows enlargement of main pulmonary artery consistent with pulmonary hypertension, but no evidence of pulmonary embolism The

presence of pulmonary hypertension in the absence of pulmonary embolism or lung disease suggests AIDS-related pulmonary hypertension with plexogenic arteriopathy.

A

B

C

D

reconstructions may provide additional information in

selected cases (20), largely in regard to lung disease

dis-tribution, but routine transverse images are adequate for

diagnosis in the large majority of cases

It has been suggested that the use of 2D coronal

re-constructions may be useful for the primary

interpreta-tion of thoracic CT, but at present, it would seem most

appropriate to use multiplanar reconstructions as a

com-pliment to axial images Kwan et al (57) compared the

accuracy and efficiency of primary interpretation of

tho-racic MDCT (5-mm slice thickness) using coronal

refor-mations to that of routine transverse images Each image

set was assessed for 58 abnormalities of the lungs,

me-diastinum, pleura, chest wall, diaphragm, abdomen, and

skeleton The mean detection sensitivity of all lesions was

significantly (p = 0.001) lower on coronal (44% ± 26%

[SD]) than on transverse (51% ± 22%) images, whereas

the mean detection specificity was significantly (p = 0.005) higher (96% ± 5% vs 95% ± 6%, respectively) Also, reporting findings for fewer coronal images took

significantly (p = 0.025) longer (mean, 263 ± 56 s vs 238

± 45 s, respectively) (57) Arakawa et al (58) evaluated the diagnostic utility of coronal MD-HRCT reformations (1.9-mm thickness) to axial HRCT (2-mm collimation)

in diffuse and focal lung diseases In 22.1% of cases, coronal MD-HRCT reformations were regarded as supe-rior to axial HRCT or provided additional information, whereas in 72.4%, coronal MD-HRCT was regarded

C H A P T E R 1 Technical Aspects of High-Resolution CT 17

thickness in a 19-year-old woman with hypoxemia Transaxial (A and B) images

show numerous very small subpleural arteriovenous malformations (arrows)

One-centimeter-thick MIP images in the transaxial (C, D) and coronal (E) planes show

the malformations (arrows) and their vascular supply to better advantage She was

subsequently found to have Osler-Weber-Rendu disease.

FIGURE 1-20 Coronal (A) and sagittal (B) MinIP reconstruction from 1.25-mm MD-HRCT in a patient with lymphangiomyomatosis The MinIP images optimize

visualization of the lung cysts and their distribution.

A: A single HRCT image shows two small nodules (arrows) that are difficult to distinguish from vessels B: An MIP image consisting of eight contiguous HRCT images, including

A, allows the two small nodules to be easily distinguished from surrounding vessels.

as comparable to axial HRCT, and in 5.5% it was

con-sidered inferior to axial images (58) Remy- Jardin et al

(59) assessed the diagnostic accuracy of coronal

recon-structions as an alternative to transverse MD-HRCT in

the diagnosis of infiltrative lung disease No significant

difference was found between the transverse and coronal images in the identification of CT features of disease or their distribution in the central, peripheral, anterior, and/

or posterior lung zones However, in patients with sive lung disease, the cephalocaudal distribution of lung

C H A P T E R 1 Technical Aspects of High-Resolution CT 19

same anatomical level Normal lung parenchyma appears relatively homogeneous

Pulmonary vessels disappear on MinIP images.

HRCT image shows a typical patchy distribution of interlobular septal thickening and ground-glass opacity (i.e., crazy paving) typical of alveolar proteinosis B: A MIP image

consisting of five contiguous HRCT images, including A, results in a confusing superimposition of opacities Septal thickening is more difficult to diagnose.

abnormalities was more precisely assessed with coronal

reconstructions (59)

Nishino et al (60) attempted to determine whether

sagittal reformations of volumetric MD-HRCT provide

additional information in evaluating lung

abnormali-ties, when compared to axial HRCT images Additional

findings of diagnostic significance were identified on the

sagittal reconstructions in 2 or 22 patients, principally

related to the relationship of a nodule or mass to the

fis-sures, pleura, or pericardium (60)

Maximum- and Minimum-Intensity Projections

Several studies have used helical HRCT with thin

collima-tion and MIPs or MinIPs to acquire and display

volumet-ric HRCT data for a slab of lung (20,61–63) In a study

by Bhalla et al (61), when compared to conventional

HRCT, volumetric MIP and MinIP images demonstrated

additional findings in 13 of 20 (65%) cases However,

the authors found that the conventional HRCT scans

showed fine linear structures, such as the walls of airways and interlobular septa, more clearly than either MIP or MinIP images

MIP imaging has been used to best advantage for the diagnosis of nodular lung disease MIP images increase the detection of small lung nodules and can be helpful in demonstrating their anatomical distribution (Fig 1-21) Coakley et al (62) assessed the use of MIP images in the detection of pulmonary nodules by helical CT In this study, 40 pulmonary nodules of high density were created

by placement of 2- and 4-mm beads into the peripheral airways of five dogs MIP images were generated from overlapped slabs of seven consecutive 3-mm slices, recon-structed at 2-mm intervals, and acquired at pitch 2 MIP imaging increased the odds of nodule detection by more than two, when compared to helical images, and reader confidence for nodule detection was significantly higher with MIP images

In a study by Bhalla et al (61), the use of helical HRCT and MIP images was compared in patients with nodular lung disease Because of the markedly improved visual-ization of peripheral pulmonary vessels and improved spatial orientation, MIP images were considered superior

to helical scans for identifying pulmonary nodules and specifying their location as peribronchovascular or centri-lobular, a finding of great value in differential diagnosis

In another study (63), sliding-thin-slab MIP structions were used in 81 patients with a variety of lung diseases associated with small nodules In this study, pa-tients were studied using 1- and 8-mm-thick conventional

recon-CT and helical recon-CT with production of 3-, 5-, and thick MIP reconstructions When conventional CT find-ings were normal, MIPs did not demonstrate additional abnormalities When conventional CT findings were inconclusive, MIP enabled the detection of micronodules (i.e., nodules 7 mm or less in diameter) involving less than 25% of the lung When conventional CT scans showed micronodules, MIP showed the extent and distribution of micronodules and associated bronchiolar abnormalities

8-mm-FIGURE 1-24 MD-HRCT image with contrast enhancement in a patient with bronchiolitis obliterans and a clinical suspicion of pulmonary embolism No pulmonary embolism was found A: A single HRCT image with 1.25-mm detector

width shows bronchiectasis and patchy lung attenuation with reduced artery size

in lucent lung regions due to air trapping and mosaic perfusion B: A 10-mm-thick

MinIP image at the same level as A accentuates the differences in attenuation

between normal lung and lucent lung, but pulmonary arteries cannot be assessed Bronchiectasis is well seen using MinIP imaging C: MIP at the same level as

B shows reduced vessel size in the lucent lung regions Inhomogeneous lung

attenuation is also visible The bronchiectasis is difficult to see on the MIP image.

C

to better advantage The sensitivity of MIP (3-mm-thick

MIP, 94%; 5-mm-thick MIP, 100%; 8-mm-thick MIP,

92%) was significantly higher than that of conventional

CT (8-mm-thick, 57%; 1-mm-thick, 73%) in the

de-tection of micronodules (p < 0.001) The authors (63)

concluded that sliding-thin-slab MIPs may help detect

micronodular lung disease of limited extent and may be

considered a valuable tool in the evaluation of diffuse

in-filtrative lung disease

Sakai et al attempted to determine whether MIP

im-ages assisted in the diagnosis of the distribution of

mi-cronodules in a variety of focal and diffuse infiltrative lung

diseases Ten-millimeter-thick MIP image slabs at 10-mm

intervals were produced from MD-HRCT Radiology

res-idents interpreting the images benefited significantly from

the use of MIPs, while board-certified radiologists had

equal accuracy with and without the MIPs (64)

Although MIP imaging may be valuable in the

detec-tion and diagnosis of lung nodules or nodular lung

dis-ease and in the demonstration of vascular abnormalities

(Fig.  1-19), in patients with other abnormalities, MIP

imaging may result in a confusing superimposition of

opacities that tends to obscure anatomical detail This is

particularly true in patients with extensive ground-glass

opacity or reticular opacities (Figs 1-16D and 1-22)

The utility of MinIP images (Figs 1-20, 1-23, and 1-24)

has also been evaluated MinIP images are most useful in

the demonstration of abnormalities characterized by low attenuation (Figs 1-20 and 1-24) (61) In one study (61), MinIP images were more accurate than routine HRCT scans in identifying (a) the lumina of central airways (Figs 1-16E,F and 1-24B), (b) areas of abnormal low at-tenuation (e.g., emphysema or air trapping) (Figs.  1-20 and 1-24B), and (c) ground-glass opacity

Effective dose is a widely used measure of radiation

exposure from medical imaging (70) Effective dose is calculated by summing the absorbed doses to individual organs weighted for their radiation sensitivity; the unit

of measurement is the sievert or millisievert Determining the effective dose requires the measurement of absorbed

C H A P T E R 1 Technical Aspects of High-Resolution CT 21

dose to each body organ multiplied by their radiation

sensitivity, which is impractical in the clinical setting

However, a simpler calculation may be made in order

to estimate the effective dose, based on several

assump-tions (70) Scanner manufacturers use dose data derived

from measurements of radiation dose in phantoms to

de-termine a weighted CT dose index (CTDI) for each CT

scanner model, at all available selections of tube voltage

(kV(p)), tube current (mA), and rotation time The

se-lected pitch value is then incorporated to produce a CT

dose index called the CTDIVOL, measured in grays (Gy)

or milligrays (mGy) The CTDIVOL allows a

compari-son of the amount of radiation associated with different

scanners and scan parameters, but does not take into

ac-count the length of the scan or the radiation sensitivity of

affected tissues and organs

The CTDIVOL is multiplied by the scan length in

cen-timeters to calculate the dose length product (DLP) The

DLP is a measure of the overall radiation dose delivered

to the patient during the scan An estimated effective dose

for a specified CT scan can be calculated by multiplying

the DLP by a normalized effective dose coefficient for the

scanned body part (chest = 0.014 mSv/mGy/cm or 1.4%)

(71) The effective dose coefficient accounts for the

radia-tion sensitivity of the body region scanned, although

spe-cific organs are not considered, and several assumptions

are made; tissue-weighting factors are averaged over sex

and age and patient size, or in other words, this value

assumes an average patient (70) Although it does not

provide a precise measurement, it allows a general

com-parison of imaging studies Yearly background radiation

is approximately 2.5 to 3 mSv

HRCT performed with spaced axial images results in a

low-radiation dose as compared to MD-HRCT obtained

with volumetric image acquisition (Fig 1-25, Table 1-2)

(16,72) For example, Mayo et al (72) compared the

tho-racic radiation dose associated with spaced axial HRCT

to that of conventional CT with contiguous slices In this

study, using an early-generation scanner and scan

tech-nique of 120 kV(p), 200 mA, and 2-s scan time, the mean

skin radiation dose was 4.4 mGy for 1.5-mm HRCT

scans at 10-mm intervals, 2.1 mGy for scans at 20-mm

intervals, and 36.3 mGy for conventional 10-mm scans

at 10-mm intervals Thus, HRCT scanning at 10- and

20-mm intervals, as done in clinical imaging, resulted in

12% and 6%, respectively, of the radiation dose

associ-ated with conventional CT Schoepf et al (49) compared

MD-HRCT to HRCT obtained with spaced axial images

considered to have equal image quality, spatial

resolu-tion, subjective signal-to-noise ratio, diagnostic value,

depiction of bronchi and parenchyma, and motion and

streak artifacts Radiation dose measured 5.55 mSv for

MDCT and 1.25 mSv for the series of 24 axial HRCT

slices obtained (49) Leswick et al found that if image

noise is equalized, MD-HRCT may result in a higher

ra-diation dose than the combination of a routine MDCT

sufficient for volumetric imaging and spaced axial HRCT

images (37)

HRCT images were obtained in the supine (A) and prone (B) positions at 1-cm

intervals using a modulated tube current varying between 100 and 150 mA Dynamic expiratory images (C) were also obtained at three selected levels using a tube current

(mA) of 50 Estimated effective dose for the examination was 1.5 mSv.

A

B

C

Procedure Effective radiation dose (mSv)

Annual background radiation 2.5

PA chest radiograph 0.05 Spaced axial HRCT (10-mm spacing) 0.7 Spaced axial HRCT, supine, prone

(10-mm spacing), expiratory 1.5 Spaced axial HRCT (20-mm spacing) 0.35 Low-dose spaced axial HRCT 0.02 MD-HRCT (standard technique) 4–7 MD-HRCT (modulated mA of approx 100) 2–3

Modified from Mayo JR, Aldrich J, Muller NL Radiation exposure at chest CT:

a statement of the Fleischner Society Radiology 2003;228:15–21.

TABLE 1-2 Comparison of Radiation Dose for Chest Imaging Techniques

Attempts at dose reduction with MD-HRCT using a

decrease in mAs may be achieved by choosing a reduced

fixed mA value, body weight-based formulas, or

scanner-based dynamic tube current modulation (19,70,73,74)

Tube current modulation can provide excellent HRCT

studies with reduced radiation dose (Fig 1-26)

However, as pointed out by Mayo et al (16,75), dose

reduction may have an adverse effect on image quality

and reader interpretations For example, Yi et al (76) assessed image noise and subjective image quality with respect to the radiation dose delivered by MDCT in

20 patients with suspected bronchiectasis Images were obtained using 120 kV(p), 2.5-mm collimation, pitch of 1.5, 2.5-mm reconstruction intervals, and sharp recon-struction algorithm The quality of the images obtained using six mA settings (170, 100, 70, 40, 20, and 10 mA) was assessed, and it was graded using a 5-point scale (5 = excellent to 1 = nondiagnostic) at both lung and mediastinal window settings Also, radiation doses were measured at each of the six mA settings using a thoracic phantom The mean image quality scores at exposures

of 170, 100, 70, 40, 20, and 10 mA were 3.9, 3.7, 3.8, 3.2, 2.5, and 1.6 at lung window settings, and images obtained at 70 mA were rated significantly better than

those obtained at 40  mA or less (p < 0.01) The

aver-age imaver-age noise (SD of pixels measured in blood) was 39, 42.7, 53.6, 69.2, 98.5, and 157.2 H, respectively, at 170,

100, 70, 40, 20, and 10 mA, and the mean radiation doses measured at these mA values were 23.72, 14.39, 10.54, 5.41, 2.74, and 1.50 mGy, respectively (76) The authors point out that the dose resulting from MDCT obtained with 70 mA (10.54 mGy) is five times that reported for spaced axial HRCT (2.17 mGy with parameters of 120 kV(p), 170 mA, 1-mm collimation, and 10-mm intervals) for the diagnosis of bronchiectasis (77)

Das et al (19) compared the image quality of racic MDCT obtained with a standard protocol (effective mAs = 100) to three methods of dose reduction, including

tho-a dyntho-amic tube current modultho-ation, effective mAs equtho-als body weight in kilograms, and a combination of these The mean effective doses for these protocols, respectively, were 6.83, 5.92, 4.73, and 3.97 mSv Although there was

a correlation between decreased dose and increased age noise, the image quality for all techniques was graded

im-as excellent (19)

Tube current modulation allows for dynamic changes

of the tube current (mA) in the craniocaudal (Z plane) and transaxial (X and Y) planes Tube output varies de-pending upon the attenuation profiles of specific anatom-ical locations For example, tube current will be lowered

in regions of the body that have less attenuation, such as levels at which the lungs comprise a large portion of the cross-sectional area of the chest Tube current modulation attempts to maintain fixed image noise at all anatomi-cal levels, reducing radiation exposure without sacrificing image quality Using tube current modulation, Kalra et al (78) showed a dose reduction of 18% to 26% compared

to a fixed mA in patients undergoing routine chest CT Angel et al (79) demonstrated a 16% reduction in the absorbed dose to the lung with tube current modulation,

an effect that was most pronounced in smaller patients

In larger patients, there was an increase in the dose of

up to 33% (79) When tube current modulation is used, changing other parameters such as pitch and gantry rota-tion speed will have limited impact on radiation dose For example, increasing the pitch will result in a subsequent

approximately 100 Representative images through the upper (A), mid- (B), and

lower (C) lungs are shown from a supine volumetric HRCT acquired with 120 kV(p),

a fixed tube current of 100 mA, and reconstructed at 1.25-mm thickness Dynamic

expiratory images were also obtained at three selected levels The estimated effective

dose for the examination was 2 mSv The dose would be increased by the inclusion

of prone images.

A

B

C

C H A P T E R 1 Technical Aspects of High-Resolution CT 23

elevation in tube current so that noise image remains

constant The primary exception is when tube current is

already at its maximum level, such as in large patients

With tube current modulation, and a state-of-the-art

scanner having sensitive detectors, HRCT (supine

volu-metric, prone axial at 1-cm intervals, and dynamic

expi-ratory imaging at three levels) may be performed using

300-ms rotation and an mA averaging about 100, with an

estimated dose of about 2 to 3 mSv, and excellent image

quality Supine and prone axial imaging (1-cm spacing)

and dynamic expiratory imaging at three levels can be

performed with an estimated dose of about 1 mSv

Another dose reduction strategy is the use of

adap-tive statistical iteraadap-tive reconstruction (ASIR) ASIR uses

a postprocessing algorithm that represents an adjunct to

standard filtered back projection (FBP) In usual

clini-cal practice, a combination of FBP and ASIR are used to

produce the final data set with typical blends, including

30% to 40% ASIR Reconstructions using ASIR have

re-duced image noise compared to those using only FBP,

al-lowing images to be acquired with parameters that lower

the radiation dose (70) However, the use of ASIR can

affect quantitative CT measurements (80) In one study

(81), images were acquired at various tube current-time

products (40–150 mAs) and then reconstructed using FBP

and blended ASIR/FBP At 40 and 75 mAs, the images

reconstructed with FBP had unacceptable levels of noise,

whereas the ASIR/FBP images had acceptable noise levels

(81) In the evaluation of diffuse lung disease, Prakash

et al (82) showed that images reconstructed with ASIR

in a high-definition mode were superior in quality to FBP

in 64% of cases ASIR’s primary disadvantages are an

in-creased postprocessing time (30% longer than FBP), edge

definition artifacts, and the production of oversmoothed

images These disadvantages are limited by the blending

of ASIR and FBP in the final reprocessing Model-based

iterative reconstruction (MBIR) is a more advanced form

of iterative reconstruction that allows for further

reduc-tion in radiareduc-tion dose at the expense of significantly

in-creased postprocessing time, but is not in common use at

this time

Radioprotective bismuth shields are another dose

reduction technique that allows for a decrease in the

specific target dose to radiosensitive organs such as the

breasts and thyroid Bismuth shields enable a reduction

in the target organ dose at the expense of increased

ar-tifacts With breast shields, in particular, this artifact is

most pronounced in the anterior lungs (83) In a study by

Colombo et al (84), bismuth shielding allowed for a 34%

reduction in the dose to the breast during chest CT with

only a slight degradation of image quality The

contem-poraneous use of tube current modulation and bismuth

shields may result in an increase in the tube current or

im-age noise depending upon when the shield is applied,

be-fore or after the scout image (85,86) Leswick et al (86)

showed that z-axis automatic tube current modulation

was more effective than shielding in reducing the

radia-tion exposure to the thyroid A combinaradia-tion of shielding

and automatic tube current modulation reduced the thyroid dose slightly compared to tube current modula-tion alone; however, this was at the expense of increased artifact (86) The American Association of Physicists in Medicine recommends using alternative methods of dose reduction, in lieu of shields, because of their unpredict-able effects on image quality and radiation exposure, particularly when automatic tube current modulation is used (87)

Low-Dose Axial High-Resolution Computed Tomography

Spaced axial HRCT with reduced mAs can allow a diagnosis of diffuse lung disease with very limited radia-tion exposure Obtaining spaced axial HRCT at 20-mm intervals (40 mAs) or at three levels (80 mAs) results in

an average skin dose comparable to that associated with chest radiography (72,88–91) Low-dose HRCT should not be routinely used for the initial evaluation of patients with lung disease, although it can be valuable in follow-ing patients with a known lung abnormality or in screen-ing large populations at risk for lung disease Optimal low-dose techniques will likely vary with the clinical set-ting and indication for the study, and they remain to be established

The efficacy of low-dose spaced axial HRCT has been assessed in several studies (88,89,92,93) In a study by Zwirewich et al (88), scans with 1.5-mm collimation and 2-s scan time at 120 kV(p) were obtained using both

20  mA (low-dose HRCT) and 200 mA dose HRCT) at selected levels in the chests of 31 patients Observers evaluated the visibility of normal structures, various parenchymal abnormalities, and artifacts using both techniques Low- and conventional-dose HRCT were equivalent for the demonstration of vessels, lobular and segmental bronchi, and structures of the secondary pulmonary lobule, and in characterizing the extent and distribution of reticular abnormalities, honeycomb cysts, and thickened interlobular septa However, the low-dose technique failed to demonstrate ground-glass opacity in 2

(conventional-of 10 cases, and emphysema in 1 (conventional-of 9 cases, although they were evident but subtle on the usual-dose HRCT Lin-ear streak artifacts were also more prominent on images acquired with the low-dose technique, but the two tech-niques were judged equally diagnostic in 97% of cases The authors concluded that HRCT images acquired at

20 mA yield anatomical information equivalent to that obtained with 200-mA scans in the majority of patients without significant loss of spatial resolution or image degradation due to streak artifacts

In a subsequent study (89), the diagnostic accuracies of chest radiographs, low-dose HRCT (80 mAs, 120 kV(p)), and conventional-dose HRCT (340 mAs, 120 kV(p)) were compared in 50 patients with chronic infiltrative lung disease and 10 normal controls For each HRCT technique, only three images were used, obtained at the levels of the aortic arch, tracheal carina, and 1 cm above

the right hemidiaphragm A correct first-choice

diagno-sis was made significantly more often with either HRCT

technique than with radiography; the correct diagnosis

was made in 65% of cases using radiographs, 74% of

cases with low-dose HRCT (p < 0.02), and 80% of

con-ventional HRCT (p < 0.005) A high confidence level in

making a diagnosis was reached in 42% of radiographic

examinations, 61% of the low-dose HRCT examinations

(p < 0.01), and 63% of the conventional-dose HRCT

examinations (p < 0.005), and it was correct in 92%,

90%, and 96% of the studies, respectively Although

conventional-dose HRCT was more accurate than

low-dose HRCT, this difference was not significant, and both

techniques provided quite similar anatomical information

(Figs 1-25 and 1-26) (89)

In a comparison of standard (150 mAs) and low (40

mAs)-dose thin-section volumetric chest CT, Christie

et al (94) found that there was significantly increased

de-tection of ground-glass opacities, ground-glass nodules,

and interstitial opacities with the higher-dose scan The

detection of solid nodules, airspace disease, and airways

disease was equivalent using low- and high-dose images

Majurin et al (92) compared a variety of low-dose

techniques in 45 patients with suspected asbestos-related

lung disease Of the 37 patients with CT evidence of lung

fibrosis, HRCT images obtained with mAs as low as 120

clearly showed parenchymal bands, curvilinear opacities,

and honeycombing However, reliable identification of

in-terstitial lines or areas of ground-glass opacity required a

minimum technique of 160 mAs Furthermore, these

au-thors showed that using the lowest possible dosage (60

mAs) HRCT was sufficient only for detecting marked

pleural thickening and areas of gross lung fibrosis

An additional factor in obtaining low-dose HRCT is a

consideration of the anatomical distribution of suspected

disease Significant dose reduction can be achieved by

lim-iting scanning to the most appropriate lung regions and

the most appropriate patient positions for obtaining the

scans As an example, in screening for asbestosis,

scan-ning in the prone position and the posterior lung bases is

most helpful in diagnosis (Fig 1-27)

EXPIRATORY HIGH-RESOLUTION

COMPUTED TOMOGRAPHY

As an adjunct to routine inspiratory images, expiratory

HRCT scans have proved useful in the evaluation of

pa-tients with a variety of obstructive lung diseases (95,96)

On expiratory scans, focal or diffuse air trapping may be

diagnosed in patients with large or small airway

obstruc-tion or emphysema It has been shown that the presence of

air trapping on expiratory scans (a) correlates to some

de-gree with pulmonary function test abnormalities (97,98),

(b) can confirm the presence of obstructive airway

dis-ease in patients with subtle or nonspecific abnormalities

visible on inspiratory scans, (c) allows the diagnosis of

significant lung disease in some patients with normal

in-spiratory scans (99), and (d) can help distinguish between

obstructive disease and infiltrative disease as a cause of inhomogeneous lung opacity seen on inspiratory scans (100)

In most lung regions of normal subjects, lung chyma increases uniformly in attenuation during expi-ration (8,101–105), but in the presence of air trapping, lung parenchyma remains lucent on expiration and shows little change in volume Focal, multifocal, or diffuse air trapping is visible as areas of abnormally low attenuation

paren-on expiratory or postexpiratory CT On expiratory scans, visible differences in attenuation between normal and obstructed lung regions are visible using standard lung window settings and can be quantitated using regions of interest Differences in attenuation between normal lung regions and regions that show air trapping often measure more than 100 Hounsfield units (HU) (106) Air trapping visible using expiratory or postexpiratory HRCT tech-niques has been recognized in patients with emphysema (107–110), chronic airways disease (98), asthma (111–115), cystic fibrosis (116), bronchiolitis obliterans and BOS (99,108,117–127), the cystic lung diseases associ-ated with Langerhans histiocytosis and tuberous sclerosis (128), bronchiectasis (108,129), airways disease related

to AIDS (130), and small airways disease associated with thalassemia (131) Expiratory HRCT has also proved valuable in demonstrating the presence of bronchiolitis in patients with primarily infiltrative diseases such as hyper-sensitivity pneumonitis (132,133), sarcoidosis (134–137), and pneumonia

Some investigators obtain expiratory scans routinely

in all patients who have HRCT, whereas others limit their use to patients with inspiratory scan abnormalities or suspected obstructive lung disease (95) We recommend the routine use of expiratory scans in a patient’s initial HRCT evaluation because the functional cause of respi-ratory disability is not always known before HRCT is performed Furthermore, even in patients with a known restrictive abnormality on pulmonary function tests, or obvious HRCT findings of fibrosis, expiratory HRCT

obtained at 1-cm intervals in the prone position using a fixed tube current of 40 mA

No supine or expiratory images were obtained Estimated effective dose for the examination was 0.2 mSv.

C H A P T E R 1 Technical Aspects of High-Resolution CT 25

may show air trapping, a finding of potential value in

dif-ferential diagnosis (136) For example, the presence of air

trapping in a patient with HRCT findings of fibrosis and

honeycombing excludes the diagnosis of usual

intersti-tial pneumonia and idiopathic pulmonary fibrosis (138)

Limiting expiratory HRCT to patients with evidence of

airway abnormalities on inspiratory scans will result in

some missed diagnoses Expiratory HRCT may also show

findings of air trapping in the absence of inspiratory scan

abnormalities (99) The use of expiratory scans may be

of value in the follow-up of patients at risk of

develop-ing an obstructive abnormality For example, expiratory

scans are valuable in detecting bronchiolitis

obliter-ans in patients being followed for lung trobliter-ansplantation

(123,125,139–142)

Expiratory HRCT scans may be obtained during

sus-pended respiration after forced exhalation

(postexpira-tory CT), during forced exhalation (dynamic expira(postexpira-tory

CT) (95,104,108,143), at a user-selected respiratory level

controlled during exhalation with a spirometer

(spiro-metrically triggered expiratory CT) or by using other

methods (126,144–149) Generally, with these

tech-niques, expiratory scans are obtained at selected levels

Three scans, five scans, or scans at 4-cm intervals have

been used by different authors Expiratory imaging may

also be performed using helical technique and 3D

volu-metric reconstruction (150,151)

Postexpiratory High-Resolution

Computed Tomography

Postexpiratory HRCT scans, obtained during suspended

respiration after a forced exhalation, are easily

per-formed with any scanner and are most suitable for a

routine examination (Fig 1-28) The primary advantage

of this technique is its simplicity In obtaining

expira-tory HRCT, the patient is instructed to forcefully exhale

and then hold his or her breath for the duration of the

single scan This maneuver is practiced with the patient before the scans are obtained to ensure an adequate level of expiration Postexpiratory scans can be per-formed at several predetermined levels (e.g., aortic arch, carina, lung bases), at 2- to 4-cm intervals, or at levels appearing abnormal on the inspiratory images Scans at two to five levels have been used by different authors (100,111,112,120,140,152) Expiratory scans at three selected levels (aortic arch, hila, and lower lobes) are generally sufficient for showing significant air trapping and may be used routinely, in addition to the inspira-tory scan series, in patients with suspected airways or obstructive lung diseases Although targeting postexpi-ratory scans to lung regions that appear abnormal on the inspiratory scans would seem advantageous, using preselected scans allows the same lung regions to be rou-tinely imaged on follow-up examinations and, in some patients, can show air trapping when inspiratory scans are normal

Each postexpiratory scan is compared to the tory scan that most closely duplicates its level to detect air trapping Anatomical landmarks such as pulmonary vessels, bronchi, and fissures are most useful for local-izing corresponding levels Because of diaphragmatic mo-tion occurring with expiration, attempting to localize the same scan levels by using the scout view is difficult and sometimes misleading

inspira-Dynamic Expiratory High-Resolution Computed Tomography

Scans obtained dynamically during forced expiration can

be obtained using an electron-beam scanner (Fig 1-29)

or a helical scanner (Figs 1-30 to 1-33) There is some evidence to suggest that a greater increase in lung attenu-ation occurs with dynamic expiratory imaging than with simple postexpiratory HRCT and that, consequently, air trapping is more easily diagnosed (Fig 1-33)

attenuation without evidence of airways disease B: Routine postexpiratory scan shows patchy air trapping (arrows) indicative of small airways disease.