Radiologic Physics Taught Through Cases 9781626239678, 9781626239715

Table of contents :
Radiologic Physics Taught Through Cases
Title Page
Copyright
Contents
Preface
Contributors
1 Fluoroscopy
Introduction
Case 1: SID, ABC, and Radiation Output
1.1.1 Background
1.1.2 Findings
1.1.3 Discussion
1.1.4 Resolution
Case 2: Reference Air Kerma and Skin Dose
1.2.1 Background
1.2.2 Findings
1.2.3 Discussion
1.2.4 Resolution
Case 3: Collimation
1.3.1 Background
1.3.2 Findings
1.3.3 Discussion
1.3.4 Resolution
Case 4: Anti-scatter Grids
1.4.1 Background
1.4.2 Findings
1.4.3 Discussion
1.4.4 Resolution
Case 5: Patient Shielding
1.5.1 Background
1.5.2 Findings
1.5.3 Discussion
1.5.4 Resolution
Case 6: CT Fluoroscopy
1.6.1 Background
1.6.2 Findings
1.6.3 Discussion
1.6.4 Resolution
Case 7: Digital Subtraction Angiography and Motion Artifacts
1.7.1 Background
1.7.2 Findings
1.7.3 Discussion
1.7.4 Resolution
Case 8: Fluoroscopy Modes and Dose
1.8.1 Background
1.8.2 Findings
1.8.3 Discussion
1.8.4 Resolution
Case 9: Equalization Filters
1.9.1 Background
1.9.2 Findings
1.9.3 Discussion
1.9.4 Resolution
Case 10: Cone Beam Computed Tomography
1.10.1 Background
1.10.2 Findings
1.10.3 Discussion
1.10.4 Resolution
Review Questions
1.11.1 Case 1: SID, ABC, and Radiation Output
1.11.2 Case 2: Reference Air Kerma and Skin Dose
1.11.3 Case 3: Collimation
1.11.4 Case 4: Anti-scatter Grids
1.11.5 Case 5: Patient Shielding
1.11.6 Case 6: CT Fluoroscopy
1.11.7 Case 7: Digital Subtraction Angiography and Motion Artifacts
1.11.8 Case 8: Fluoroscopy Modes and Dose
1.11.9 Case 9: Equalization Filters
1.11.10 Case 10: Cone Beam Computed Tomography
References
2 Mammography
Introduction
Common Image Quality Problems
Case 1: Magnification Imaging
2.1.1 Background
2.1.2 Findings
2.1.3 Discussion
2.1.4 Resolution
Case 2: Focal Spot Size Selection in Magnification Views
2.2.1 Background
2.2.2 Findings
2.2.3 Discussion
2.2.4 Resolution
Case 3: X-ray Acquisition Technique Factors in Mammography
2.3.1 Background
2.3.2 Findings
2.3.3 Discussion
2.3.4 Resolution
Case 4: Digital Breast Tomosynthesis: Artifacts due to High-Contrast Objects
2.4.1 Background
2.4.2 Findings
2.4.3 Discussion
2.4.4 Resolution
Case 5: Effect of Image Post-Processing on the Appearance of a Mammogram
2.5.1 Background
2.5.2 Findings
2.5.3 Discussion
2.5.4 Resolution
Case 6: Artifact due to Detector Row Dropout
2.6.1 Background
2.6.2 Findings
2.6.3 Discussion
2.6.4 Resolution
Case 7: Microcalcification-like Appearance Caused by a Detector Artifact
2.7.1 Background
2.7.2 Findings
2.7.3 Discussion
2.7.4 Resolution
Case 8: Artifact due to Imperfection in Compression Paddle
2.8.1 Background
2.8.2 Findings
2.8.3 Discussion
2.8.4 Resolution
Case 9: Patient Motion Causing Blurred Parenchymal Structure in a Mammogram
2.9.1 Background
2.9.2 Findings
2.9.3 Discussion
2.9.4 Resolution
Case 10: EMI Artifact due to LVAD Device
2.10.1 Background
2.10.2 Findings
2.10.3 Discussion
2.10.4 Resolution
Review Questions
2.11.1 Case 1: Magnification Imaging
2.11.2 Case 2: Focal Spot Size Selection in Magnification Views
2.11.3 Case 3: X-ray Acquisition Technique Factors in Mammography
2.11.4 Case 4: Digital Breast Tomosynthesis: Artifacts due to High-Contrast Objects
2.11.5 Case 5: Effect of Image Post-Processing on the Appearance of a Mammogram
2.11.6 Case 6: Artifact due to Detector Row Dropout
2.11.7 Case 7: Microcalcificationlike Appearance Caused by a Detector Artifact
2.11.8 Case 8: Artifact due to Imperfection in Compression Paddle
2.11.9 Case 9: Patient Motion Causing Blurred Parenchymal Structure in a Mammogram
2.11.10 Case 10: EMI Artifact due to LVAD Device
Equations
References
3 Computed Tomography
Introduction
Case 1: Ring Artifact
3.1.1 Background
3.1.2 Findings
3.1.3 Discussion
3.1.4 Resolution
Case 2: Effect of Patient Size on CT Number Accuracy
3.2.1 Background
3.2.2 Findings
3.2.3 Discussion
3.2.4 Resolution
Case 3: Effect of kV Selection on Image Quality and Dose
3.3.1 Background
3.3.2 Findings
3.3.3 Discussion
3.3.4 Resolution
Case 4: Image Quality Variation with Reconstructed Slice Thickness
3.4.1 Background
3.4.2 Findings
3.4.3 Discussion
3.4.4 Resolution
Case 5: Image Quality Variation with Reconstruction Filter
3.5.1 Background
3.5.2 Findings
3.5.3 Discussion
3.5.4 Resolution
Case 6: Displayed Volume CT Dose Index and Patient Size
3.6.1 Background
3.6.2 Findings
3.6.3 Discussion
3.6.4 Resolution
Case 7: Beam Hardening Artifact
3.7.1 Background
3.7.2 Findings
3.7.3 Discussion
3.7.4 Resolution
Case 8: Partial Volume Artifact
3.8.1 Background
3.8.2 Findings
3.8.3 Discussion
3.8.4 Resolution
Case 9: Metal Artifact
3.9.1 Background
3.9.2 Findings
3.9.3 Discussion
3.9.4 Resolution
Case 10: Motion Artifact
3.10.1 Background
3.10.2 Findings
3.10.3 Discussion
3.10.4 Resolution
Review Questions
3.11.1 Case 1: Ring Artifact
3.11.2 Case 2: Effect of Patient Size on CT Number Accuracy
3.11.3 Case 3: Effect of kV Selection on Image Quality and Dose
3.11.4 Case 4: Image Quality Variation with Reconstructed Slice Thickness
3.11.5 Case 5: Image Quality Variation with Reconstruction Filter
3.11.6 Case 6: Displayed Volume CT Dose Index and Patient Size
3.11.7 Case 7: Beam Hardening Artifact
3.11.8 Case 8: Partial Volume Artifact
3.11.9 Case 9: Metal Artifact
3.11.10 Case 10: Motion Artifact
Equations
References
4 Magnetic Resonance Imaging
Introduction
Common Image Quality Problems
Case 1: Appearance of Discrete Image Ghosts on Abdominal Imaging
4.1.1 Background
4.1.2 Findings
4.1.3 Discussion
4.1.4 Resolution
Case 2: AWell-Defined Area of Signal Hyperintensity Appears Bilaterally at the Level of the Internal Auditory Canal on Diffusion
4.2.1 Background
4.2.2 Findings
4.2.3 Discussion
4.2.4 Resolution
Case 3: Appearance of Extra Field-of-view Anatomy on the Inferior Portion of Sagittal 3D T2-Weighted Acquisition of the Spine
4.3.1 Background
4.3.2 Findings
4.3.3 Discussion
4.3.4 Resolution
Case 4: Precontrast, Axial 3D T1-Weighted Gradient Echo with Fat Suppression Shows Adequate Anatomical Detail, but Minor Edge Ri
4.4.1 Background
4.4.2 Findings
4.4.3 Discussion
4.4.4 Resolution
Case 5: Dark Etching Appears at the Boundary of Fat and Soft-Tissue Layers
4.5.1 Background
4.5.2 Findings
4.5.3 Discussion
4.5.4 Resolution
Case 6: Application of FatSuppressed Sequences in the Pelvis did not Reveal the Expected Contrast
4.6.1 Background
4.6.2 Findings
4.6.3 Discussion
4.6.4 Resolution
Case 7: T1-Weighted Gradient Echo of the Abdomen Shows Marked Artifact Medially on Both Coronal and Axial FOV, Obscuring Visuali
4.7.1 Background
4.7.2 Findings
4.7.3 Discussion
4.7.4 Resolution
Case 8: Abnormal Dark Fluid Seen in the Bladder of an Axial Single-Shot T2Weighted Sequence, but not on Location-Matched 3D T2 A
4.8.1 Background
4.8.2 Findings
4.8.3 Discussion
4.8.4 Resolution
Case 9: Postcontrast T1Weighted Gradient Echo Reveals Patchy Enhancement in the Anterior SeptalWall
4.9.1 Background
4.9.2 Findings
4.9.3 Discussion
4.9.4 Resolution
Case 10: Significant Signal-to-Noise Variation Across the FOV, Creating Nondiagnostic Image Quality
4.10.1 Background
4.10.2 Findings
4.10.3 Discussion
4.10.4 Resolution
Review Questions
4.11.1 Case 1: Appearance of Discrete Image Ghosts on Abdominal Imaging
4.11.2 Case 2: AWell-Defined Area of Signal Hyperintensity Appears Bilaterally at the Level of the Internal Auditory Canal on Di
4.11.3 Case 3: Appearance of Extra Field-of-view Anatomy on the Inferior Portion of Sagittal 3D T2-Weighted Acquisition of the S
4.11.4 Case 4: Precontrast, Axial 3D T1-Weighted Gradient Echo with Fat Suppression Shows Adequate Anatomical Detail, but Minor
4.11.5 Case 5: Dark Etching Appears at the Boundary of Fat and Soft-Tissue Layers
4.11.6 Case 6: Application of Fat-Suppressed Sequences in the Pelvis did not Reveal the Expected Contrast
4.11.7 Case 7: T1-Weighted Gradient Echo of the Abdomen Shows Marked Artifact Medially on Both Coronal and Axial FOV, Obscuring
4.11.8 Case 8: Abnormal Dark Fluid Seen in the Bladder of an Axial Single-Shot T2-Weighted Sequence, but not on LocationMatched
4.11.9 Case 9: Postcontrast T1Weighted Gradient Echo Reveals Patchy Enhancement in the Anterior SeptalWall
4.11.10 Case 10: Significant Signal-to-Noise Variation Across the FOV, Creating Nondiagnostic Image Quality
Equations
Glossary
Suggested Reading
5 Nuclear Medicine
Introduction
Common Image Quality Problems
Case 1: Degraded Resolution of a Whole-Body Planar
Methylene Diphosphonate Image
5.1.1 Background
5.1.2 Findings
5.1.3 Discussion
5.1.4 Resolution
Case 2: Effect of Positron Range on Image Quality and Resolution
5.2.1 Background
5.2.2 Findings
5.2.3 Discussion
5.2.4 Resolution
Case 3: Standardized Uptake Value in Positron Emission Tomography (Noise Bias)
5.3.1 Background
5.3.2 Findings
5.3.3 Discussion
5.3.4 Resolution
Case 4: The Impact of Attenuation Correction in PET
5.4.1 Background
5.4.2 Findings
5.4.3 Discussion
5.4.4 Resolution
Case 5: Iterative Reconstruction and Choosing the Number of Iterations and Subsets
5.5.1 Background
5.5.2 Findings
5.5.3 Discussion
5.5.4 Resolution
Case 6: The Effects of Image Smoothing
5.6.1 Background
5.6.2 Findings
5.6.3 Discussion
5.6.4 Resolution
Case 7: Choosing the Correct Acquisition Image Matrix Size
5.7.1 Background
5.7.2 Findings
5.7.3 Discussion
5.7.4 Resolution
Case 8: Assessing Patient Motion in Myocardial Perfusion Imaging
5.8.1 Background
5.8.2 Findings
5.8.3 Discussion
5.8.4 Resolution
Case 9: Bremsstrahlung Imaging of
Microspheres Liver Embolization
5.9.1 Background
5.9.2 Findings
5.9.3 Discussion
5.9.4 Resolution
Case 10: Degraded Image Quality from an Improper Collimator
5.10.1 Background
5.10.2 Findings
5.10.3 Discussion
5.10.4 Resolution
Review Questions
5.11.1 Case 1: Degraded Resolution of a Whole-Body Planar
Methylene Diphosphonate Image
5.11.2 Case 2: Effect of Positron Range on Image Quality and Resolution
5.11.3 Case 3: Standardized Uptake Value in Positron Emission Tomography (Resolution and Noise)
5.11.4 Case 4: The Impact of Attenuation Correction in PET
5.11.5 Case 5: Iterative Reconstruction and Choosing the Number of Iterations and Subsets
5.11.6 Case 6: The Effects of Image Smoothing
5.11.7 Case 7: Choosing the Correct Acquisition Image Matrix Size
5.11.8 Case 8: Assessing Patient Motion in Myocardial Perfusion Imaging
5.11.9 Case 9: Bremsstrahlung Imaging of
Microspheres Liver Embolization
5.11.10 Case 10 Degraded Image Quality from an Improper Collimator
References
6 Ultrasound Imaging
Introduction
Common Image Quality Problems
Case 1: Pulse-Echo Imaging Principle and Speed of Sound Propagation
6.1.1 Background
6.1.2 Findings
6.1.3 Discussion
6.1.4 Resolution
Case 2: Array Transducers and Sound Frequency
6.2.1 Background
6.2.2 Findings
Linear Array
Curvilinear or Curved Array
Phased Array
6.2.3 Discussion
6.2.4 Resolution
Case 3: Nonuniformity (Array Transducer Element Dropouts)
6.3.1 Background
6.3.2 Findings
6.3.3 Discussion
6.3.4 Resolution
Case 4: Pulse-Echo Imaging Acquisition Controls
6.4.1 Background
6.4.2 Findings
6.4.3 Discussion
6.4.4 Resolution
Case 5: Reflection (Boundary Conditions)? Reverberation Artifacts
6.5.1 Background
6.5.2 Findings
6.5.3 Discussion
6.5.4 Resolution
Case 6: Range Ambiguity in B-Mode
6.6.1 Background
6.6.2 Findings
6.6.3 Discussion
6.6.4 Resolution
Case 7: Shadowing and Enhancement (Increased Through Transmission)
6.7.1 Background
6.7.2 Findings
6.7.3 Discussion
6.7.4 Resolution
Case 8: Harmonic Imaging
6.8.1 Background
6.8.2 Findings
6.8.3 Discussion
6.8.4 Resolution
Case 9: Ultrasound Image Display on Scanners and in Reading Rooms
6.9.1 Background
6.9.2 Findings
6.9.3 Discussion
6.9.4 Resolution
Case 10: Doppler Ultrasound Aliasing
6.10.1 Background
6.10.2 Findings
6.10.3 Discussion
6.10.4 Resolution
Review Questions
6.11.1 Case 1: Pulse-Echo Imaging Principle and Speed of Sound Propagation
6.11.2 Case 2: Array Transducers and Sound Frequency
6.11.3 Case 3: Nonuniformity (Array Transducer Element Dropouts)
6.11.4 Case 4: Pulse-Echo Imaging Acquisition Controls
6.11.5 Case 5: Reflection (Boundary Conditions)? Reverberation Artifacts
6.11.6 Case 6: Range Ambiguity in B-Mode
6.11.7 Case 7: Shadowing and Enhancement (Increased Through Transmission)
6.11.8 Case 8: Harmonic Imaging
6.11.9 Case 9: Ultrasound Image Display on Scanners and in Reading Rooms
6.11.10 Case 10: Doppler Ultrasound Aliasing
References
7 Image Processing
Introduction
Case 1: Filtering and Edge Enhancement
7.1.1 Background
7.1.2 Findings
7.1.3 Discussion
7.1.4 Resolution
Case 2: Maximum Intensity Projection
7.2.1 Background
7.2.2 Findings
7.2.3 Discussion
7.2.4 Resolution
Case 3: Fused Image Display of Multiple Modalities
7.3.1 Background
7.3.2 Findings
7.3.3 Discussion
7.3.4 Resolution
Case 4: Multimodality Image Registration
7.4.1 Background
7.4.2 Findings
7.4.3 Background
7.4.4 Resolution
References
Answer Key
Index

Citation preview

Radiologic Physics Taught Through Cases

Jonathon A. Nye, PhD Associate Professor Department of Radiology and Imaging Sciences Emory University School of Medicine Atlanta, Georgia

231 illustrations

Thieme New York • Stuttgart • Delhi • Rio de Janeiro

Library of Congress Cataloging-in-Publication Data is available with the publisher

© 2020. Thieme. All rights reserved. Thieme Publishers New York 333 Seventh Avenue, New York, NY 10001 USA +1 800 782 3488, [email protected] Georg Thieme Verlag KG Rüdigerstrasse 14, 70469 Stuttgart, Germany +49 [0]711 8931 421, [email protected]

Important note: Medicine is an ever-changing science undergoing continual development. Research and clinical experience are continually expanding our knowledge, in particular our knowledge of proper treatment and drug therapy. Insofar as this book mentions any dosage or application, readers may rest assured that the authors, editors, and publishers have made every effort to ensure that such references are in accordance with the state of knowledge at the time of production of the book. Nevertheless, this does not involve, imply, or express any guarantee or responsibility on the part of the publishers in respect to any dosage instructions and forms of applications stated in the book. Every user is requested to examine carefully the manufacturers’ leaflets accompanying each drug and to check, if necessary in consultation with a physician or specialist, whether the dosage schedules mentioned therein or the contraindications stated by the manufacturers differ from the statements made in the present book. Such examination is particularly important with drugs that are either rarely used or have been newly released on the market. Every dosage schedule or every form of application used is entirely at the user’s own risk and responsibility. The authors and publishers request every user to report to the publishers any discrepancies or inaccuracies noticed. If errors in this work are found after publication, errata will be posted at www.thieme.com on the product description page. Some of the product names, patents, and registered designs referred to in this book are in fact registered trademarks or proprietary names even though specific reference to this fact is not always made in the text. Therefore, the appearance of a name without designation as proprietary is not to be construed as a representation by the publisher that it is in the public domain.

Thieme Publishers Delhi A-12, Second Floor, Sector-2, Noida-201301 Uttar Pradesh, India +91 120 45 566 00, [email protected] Thieme Publishers Rio de Janeiro, Thieme Publicações Ltda. Edifício Rodolpho de Paoli, 25º andar Av. Nilo Peçanha, 50 – Sala 2508 Rio de Janeiro 20020-906 Brasil +55 21 3172 2297 Cover design: Thieme Publishing Group Typesetting by DiTech Process Solutions, India Printed in USA by King Printing Company, Inc. ISBN 978-1-62623-967-8 Also available as an e-book: eISBN 978-1-626-23971-5

54321

This book, including all parts thereof, is legally protected by copyright. Any use, exploitation, or commercialization outside the narrow limits set by copyright legislation without the publisher’s consent is illegal and liable to prosecution. This applies in particular to photostat reproduction, copying, mimeographing or duplication of any kind, translating, preparation of microfilms, and electronic data processing and storage.

Contents Preface

1.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

x

Fluoroscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

Rebecca Milman Marsh and Michael Silosky

1.1

Case 1: SID, ABC, and Radiation Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.7

Case 7: Digital Subtraction Angiography and Motion Artifacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.2

Case 2: Reference Air Kerma and Skin Dose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.8

Case 8: Fluoroscopy Modes and Dose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

1.3

Case 3: Collimation . . . . . . . . . . . . . . . . . . . 6

1.9

Case 9: Equalization Filters . . . . . . . . . . 18

1.4

Case 4: Anti-scatter Grids . . . . . . . . . . . . 8

1.10

Case 10: Cone Beam Computed Tomography . . . . . . . . . . . . . . . . . . . . . . . . . 20

1.5

Case 5: Patient Shielding . . . . . . . . . . . . 10

1.6

Case 6: CT Fluoroscopy . . . . . . . . . . . . . . 12

1.11

Review Questions . . . . . . . . . . . . . . . . . . . . 22

2.

Mammography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

25

Ingrid S. Reiser

2.1

Case 1: Magnification Imaging. . . . . . 26

2.2

Case 2: Focal Spot Size Selection in Magnification Views . . . . . . . . . . . . . . 27

2.3

Case 3: X-ray Acquisition Technique Factors in Mammography . . . . . . . . . . . 29

2.4

Case 4: Digital Breast Tomosynthesis: Artifacts due to High-Contrast Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

2.5

Case 5: Effect of Image Post-Processing on the Appearance of a Mammogram . . . . . . . . . . . . . . . . . . . 32

2.6

Case 6: Artifact due to Detector Row Dropout . . . . . . . . . . . . . . . . . . . . . . . . . 35

2.7

Case 7: Microcalcification-like Appearance Caused by a Detector Artifact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

2.8

Case 8: Artifact due to Imperfection in Compression Paddle . . . . . . . . . . . . . . 37

2.9

Case 9: Patient Motion Causing Blurred Parenchymal Structure in a Mammogram. . . . . . . . . . . . . . . . . . . . 38

2.10

Case 10: EMI Artifact due to LVAD Device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

2.11

Review Questions . . . . . . . . . . . . . . . . . . . . 40

Contents

3.

Computed Tomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

43

Karen L. Brown and Jason R. Gold

3.1

Case 1: Ring Artifact . . . . . . . . . . . . . . . . . 44

3.6

Case 6: Displayed Volume CT Dose Index and Patient Size . . . . . . . . . . . . . . . . 51

3.2

Case 2: Effect of Patient Size on CT Number Accuracy . . . . . . . . . . . . . 45

3.7

Case 7: Beam Hardening Artifact . . . 52

3.8

Case 8: Partial Volume Artifact . . . . . 53

3.3

Case 3: Effect of kV Selection on Image Quality and Dose . . . . . . . . . 46

3.9

Case 9: Metal Artifact . . . . . . . . . . . . . . . 54

3.4

Case 4: Image Quality Variation with Reconstructed Slice Thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

3.10

Case 10: Motion Artifact . . . . . . . . . . . . 55

3.11

Review Questions . . . . . . . . . . . . . . . . . . . . 56

3.5

Case 5: Image Quality Variation with Reconstruction Filter . . . . . . . . . . 49

4.

Magnetic Resonance Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

59

Puneet Sharma

4.1

Case 1: Appearance of Discrete Image Ghosts on Abdominal Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

4.6

Case 6: Application of Fat-Suppressed Sequences in the Pelvis did not Reveal the Expected Contrast . . . . . . . . 71

4.2

Case 2: A Well-Defined Area of Signal Hyperintensity Appears Bilaterally at the Level of the Internal Auditory Canal on Diffusion-Weighted MRI, Affecting Visualization of Surrounding Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

4.7

Case 7: T1-Weighted Gradient Echo of the Abdomen Shows Marked Artifact Medially on Both Coronal and Axial FOV, Obscuring Visualization of Soft Tissues . . . . . . . . 74

4.8

Case 8: Abnormal Dark Fluid Seen in the Bladder of an Axial Single-Shot T2-Weighted Sequence, but not on Location-Matched 3D T2 Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

4.9

Case 9: Postcontrast T1-Weighted Gradient Echo Reveals Patchy Enhancement in the Anterior Septal Wall . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

4.10

Case 10: Significant Signal-to-Noise Variation Across the FOV, Creating Nondiagnostic Image Quality . . . . . . . 82

4.11

Review Questions . . . . . . . . . . . . . . . . . . . . 85

4.3

4.4

4.5

vi

Case 3: Appearance of Extra Field-of-view Anatomy on the Inferior Portion of Sagittal 3D T2-Weighted Acquisition of the Spine. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Case 4: Precontrast, Axial 3D T1-Weighted Gradient Echo with Fat Suppression Shows Adequate Anatomical Detail, but Minor Edge Ripple and Blur that is Presumed to be Motion . . . . . . . . . . . . . . . . . . . . . . . . . 67 Case 5: Dark Etching Appears at the Boundary of Fat and Soft-Tissue Layers . . . . . . . . . . . . . . . . . . . . 69

Contents

5.

Nuclear Medicine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

90

Jonathon A. Nye, James R. Galt, and John N. Aarsvold

5.1

Case 1: Degraded Resolution of a Whole-Body Planar 99mTc Methylene Diphosphonate Image . . . . . . . . . . . . . . . . . . . 91

5.2

Case 2: Effect of Positron Range on Image Quality and Resolution . . . 94

5.3

5.4 5.5

6.

5.6

Case 6: The Effects of Image Smoothing . . . . . . . . . . . . . . . . . . . . . . . . . . 104

5.7

Case 7: Choosing the Correct Acquisition Image Matrix Size . . . . . 106

5.8

Case 8: Assessing Patient Motion in Myocardial Perfusion Imaging . . 108

5.9

Case 4: The Impact of Attenuation Correction in PET . . . . . . . . . . . . . . . . . . . . 99

Case 9: Bremsstrahlung Imaging of 90Y Microspheres Liver Embolization . . . . . . . . . . . . . . . . . . . . . . . . 110

5.10

Case 5: Iterative Reconstruction and Choosing the Number of Iterations and Subsets . . . . . . . . . . . . . . 101

Case 10: Degraded Image Quality from an Improper Collimator . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

5.11

Review Questions . . . . . . . . . . . . . . . . . . . 114

Case 3: Standardized Uptake Value in Positron Emission Tomography (Noise Bias) . . . . . . . . . . . . 97

Ultrasound Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

117

Zheng Feng Lu

6.1

Case 1: Pulse-Echo Imaging Principle and Speed of Sound Propagation . . . . . . . . . . . . . . . . . . . . . . . . . 118

6.2

Case 2: Array Transducers and Sound Frequency . . . . . . . . . . . . . . 120

6.3

Case 3: Nonuniformity (Array Transducer Element Dropouts) . . . . . . . . . . . . . . . . . . . . . . . . . . . 122

6.4

Case 4: Pulse-Echo Imaging Acquisition Controls . . . . . . . . . . . . . . . . 124

6.5

Case 5: Reflection (Boundary Conditions)—Reverberation Artifacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

6.6

Case 6: Range Ambiguity in B-Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

6.7

Case 7: Shadowing and Enhancement (Increased Through Transmission) . . . . . . . . . . . . . . . . . . . . . . . 130

6.8

Case 8: Harmonic Imaging . . . . . . . . . 132

6.9

Case 9: Ultrasound Image Display on Scanners and in Reading Rooms. . . . . . . . . . . . . . . . . . . 133

6.10

Case 10: Doppler Ultrasound Aliasing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

6.11

Review Questions . . . . . . . . . . . . . . . . . . . 138

vii

Contents

7.

Image Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

143

Jonathon A. Nye and Randahl C. Palmer

viii

7.1

Case 1: Filtering and Edge Enhancement . . . . . . . . . . . . . . . . . . . . . . . 144

7.3

Case 3: Fused Image Display of Multiple Modalities . . . . . . . . . . . . . . 148

7.2

Case 2: Maximum Intensity Projection . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

7.4

Case 4: Multimodality Image Registration . . . . . . . . . . . . . . . . . . . . . . . . . 150

Answer Key . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

152

Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

155

Preface Radiology residents gather their medical physics knowledge from multiple sources, often beginning with their first encounter with a radiologic image. Although many educational approaches start with fundamental physical concepts and work toward image generation, they are likely to require an extended period of time to build a conceptual framework. Arguably, the clinical demands of residency training do not always allow for a traditional classroom approach, as it can be much more efficient to learn about radiologic imaging principles during the course of a clinical rotation. Therefore, a hybrid approach may be more amenable, which begins with a study of images commonly encountered during diagnostic radiology training and provides a straightforward and compact explanation of the physical factors underlying the creation and displayed contrast of these images. To that end, this book presents a number of common physical concepts in diagnostic radiology, which may be encountered by a resident at

their clinical workstation. Although the breadth of this topic area is large, the goal of this text is to provide examples relevant to diagnostic radiology training, thereby proving to be of high value to the learner. Chapters are divided according to modality, each having 10 topics presented in a case format that is meant to quickly convey information with an image followed by a brief explanation. Some important topics, such as radiation safety, do not lend themselves to teaching from images but are part of important encounters like occupational or patient hazards during a fluoroscopic-guided procedure. In these cases, schematics are provided to assist in teaching. The reader is encouraged to consult the chapter references for further discussion. Review questions are provided at the end of each chapter to reinforce the case concepts. Jonathon A. Nye, PhD

ix

Contributors John N. Aarsvold, PhD Associate Professor Department of Radiology and Imaging Sciences Emory University School of Medicine Atlanta, Georgia

Jonathon A. Nye, PhD Associate Professor Department of Radiology and Imaging Sciences Emory University School of Medicine Atlanta, Georgia

Karen L. Brown, MHP, CHP, DABR Diagnostic Imaging Physicist Department of Radiology Penn State College of Medicine Hershey, Pennsylvania

Randahl C. Palmer, MS Resident, Department of Radiology and Imaging Sciences Emory University School of Medicine Atlanta, Georgia

James R. Galt, PhD Professor Department of Radiology and Imaging Sciences Emory University School of Medicine Atlanta, Georgia

Ingrid S. Reiser, PhD Associate Professor Department of Radiology The University of Chicago Chicago, Illinois

Jason R. Gold, DO Resident Department of Radiology Penn State Milton S. Hershey Medical Center Hershey, Pennsylvania

Puneet Sharma, PhD Assistant Professor Department of Radiology and Imaging Sciences Emory University School of Medicine Atlanta, Georgia

Zheng Feng Lu, PhD Professor Department of Radiology The University of Chicago Chicago, Illinois

Michael Silosky, MS Assistant Professor Department of Radiology University of Colorado School of Medicine Aurora, Colorado

Rebecca Milman Marsh, PhD Associate Professor Department of Radiology University of Colorado School of Medicine Aurora, Colorado

x

| 09.11.19 - 01:52

1 Fluoroscopy Rebecca Milman Marsh and Michael Silosky

Introduction Shortly after the discovery of X-rays in the late nineteenth century, fluoroscopy was developed to enable visualization of moving anatomy. Today, fluoroscopy is used for diagnosis and guidance of clinical procedures. Applications of fluoroscopy are found throughout medicine, including radiology, cardiology, urology, and speech pathology. Fluoroscopy has the same tradeoffs between image quality and radiation exposure as other X-ray-based modalities. During a fluoroscopy exam, multiple factors affect radiation dose to patients and staff. Some may be directly controlled by the fluoroscopy operator while others are dictated by patient habitus and the procedure being performed. Some interventional procedures can be very complex, requiring long fluoroscopy times to accomplish the clinical task. Fluoroscopy differs from most other imaging modalities in that the physician is often directly involved in image acquisition. In addition to

ensuring that image quality is adequate to achieve the clinical task being performed, the operator is responsible for dose management throughout the exam, not only for the safety of the patient but also for all staff members present. Understanding fluoroscopy imaging parameters and their effect on both image quality and dose is essential for anyone operating or supervising the operation of a fluoroscopy system. An essential function of modern fluoroscopy systems is the continuous adjustment of X-ray tube output to maintain adequate image quality throughout the procedure. This feedback loop between the detection system and the X-ray tube is commonly referred to as automatic brightness control (ABC), referring to the goal of maintaining a constant brightness on the output phosphor of the image intensifier.1 With the advent of digital detectors, this same concept is sometimes referred to as automatic dose control (ADC). Throughout this chapter, we will use the traditional term ABC.

1

| 09.11.19 - 01:52

Fluoroscopy

1.1 Case 1: SID, ABC, and Radiation Output 1.1.1 Background ●

A patient undergoes a fluoroscopy-guided L1

Table 1.1 Acquisition parameters, reference air kerma rates, and estimates of skin dose are shown for a range of source-to-image receptor distances

mA

Reference AKR (mGy/ minute)

Skin dose rate (mGy/ minute)

65

5.1

3.03

5.6

65

4.4

2.62

4.79

100

66

3.3

2.07

3.89

90

67

2.6

1.65

3.18

SID (cm)

kV

119 110

kyphoplasty procedure to treat vertebral body compression. ●

At the start of the procedure, the operator places the image detector 10 cm above the patient.

●

In order to properly insert the cannula into the vertebral body, the operator has to raise the

Abbreviations: AKR, air kerma rate; SID, source-to-image receptor distance.

detector to 30 cm above the patient.

1.1.2 Findings Increasing the distance between the X-ray source and the image detector increased the reported air kerma rate (AKR) by a factor of approximately 1.5.

1.1.3 Discussion Patient dose is dependent on the source-to-image receptor distance (SID) which is the distance between the source of X-rays (the focal spot of the X-ray tube) and the image receptor. As SID changes, the fluoroscopy system adjusts acquisition parameters, directly affecting the radiation output. As discussed in the beginning of this chapter, the ABC algorithms used by fluoroscopy systems adjust tube output to maintain image appearance. For a fixed X-ray output, as the distance to the image receptor increases, the amount of radiation that reaches the image receptor decreases. For example, at an SID of 100 cm, only 25% as many X-ray photons will intercept the detector compared to an SID of 50 cm. To compensate for fewer photons reaching the receptor at larger SIDs, ABC algorithms increase X-ray tube output. Typically, this is done by changing one or more of the following parameters: kV, mA, pulse width, or beam filtration. This will result in an increase in reference air kerma (AK) and, assuming a constant distance between the X-ray source and the patient’s skin, an increase in patient dose. This is illustrated in ▶ Table 1.1 where entrance dose to a phantom was measured under conditions of variable SID.

2

In addition, the distance between the X-ray source and the patient’s skin (known as the source-toskin distance [SSD]) directly impacts patient dose. Dictated by the inverse square law, doubling the distance between the radiation source and an object will result in a reduction in exposure to that object by 75%. In the case described here, the increase in X-ray tube output can be calculated using the equation: 1 ðr2 =r1 Þ2 Thus, where r1 = 50 cm and r2 = 100 cm. 1 ð100=50Þ2

¼

1 4

or 25%. When determining how the patient and equipment should be positioned for an exam, one should consider the location(s) where the operator needs to directly access the patient, how much working space is needed between the patient and the detector, any special positioning needs of the patient, and operator ergonomic factors such as table height. While some of these factors are dictated by the needs of the specific exam and operator, ensuring that SID is as small as practical can help to reduce patient dose.

1.1.4 Resolution The following steps can help reduce patient dose while maintaining affecting image quality. First, for fluoroscopic systems with fixed SID, as is the

| 09.11.19 - 01:52

1.1 Case 1: SID, ABC, and Radiation Output case with many mobile C-arms, the patient should be positioned as close to the image receptor (as far from the source) as is practical, given the needs of the procedure. Second, for fluoroscopic systems with variable SID, such as those used in interventional radiology or cardiac catheterization labs,

patient positioning may be primarily dictated by the procedure being performed. Once the patient has been positioned and the fluoroscopy system is oriented as desired, the SID should be reduced as much as is practical by moving the image receptor toward the source.

3

| 09.11.19 - 01:52

Fluoroscopy

1.2 Case 2: Reference Air Kerma and Skin Dose 1.2.1 Background ●

A patient undergoes a superior mesenteric arteriogram for embolization of a pseudoaneurysm in the transverse colon.

●

At the end of the exam, the total reference AK is 5100 mGy.

●

The case is referred to Radiation Safety so that a peak skin dose (PSD) estimate can be performed.

1.2.2 Findings The estimated PSD is 7000 mGy, approximately 1.5 times the reference AK displayed by the fluoroscopy system.

1.2.3 Discussion Fluoroscopy operators can monitor the use of radiation during procedures by paying attention to machine-reported dose metrics. All modern fluoroscopy systems are required to display the reference AK and reference AKR, where the reference AK is equivalent to the dose to air at a specific reference point. (The exact location of the reference point varies based on equipment vendor and model.) These values provide the operator with real-time dosimetry information throughout a case. Consequently, it is important to know how these values are measured and how they relate to patient exposure and patient skin dose, specifically. For fluoroscopy procedures, the primary radiation safety concern for patients is the PSD, which is the maximum dose to any single area of the skin.

It is notable that in this case, the estimated PSD is substantially greater than the AK reported by the machine. Patient skin dose is affected by several factors including table attenuation, backscatter, differences in how dose is deposited in various materials, and patient positioning. None of these factors are considered when the system calculates the displayed AK. While a thorough discussion is beyond the scope of this text, special attention should be paid to the effects of patient positioning, specifically SSD. The most common reason for large differences between the displayed AK and the actual skin dose is patient size. In the case described here, the patient was morbidly obese, and the location where the X-ray beam entered the patient’s skin was much closer to the X-ray source than the location where the fluoroscopy system calculated the AK. As a consequence, the AK at the patient’s skin was greater than the reference AK. As the AK at the patient’s skin increases, skin dose will also increase. These concepts are illustrated in ▶ Fig. 1.1. The X-ray tube is under the patient table with the X-ray focal spot (the source of the X-ray beam) indicated by a white “x.” The AK reference point (located 65 cm from the focal spot) is indicated by a black “x.” In ▶ Fig. 1.1a, the patient is closer to the X-ray tube than the AK reference point is. Here the entrance AK at the patient’s skin will be greater than the reference AK. The opposite happens in ▶ Fig. 1.1b, where the patient is farther from the source than the reference point. In this case, entrance AK will be lower than the reference AK. It should be noted that variation between the displayed AK and the entrance AK is governed by the inverse square law—as an object moves farther away from the source of radiation, the exposure decreases as a factor of the square of the Fig. 1.1 The relationship between the X-ray focal spot (white “x”), the air kerma reference point (black “x, and the entrance point of the patient’s skin are shown. (a) In this image the patient’s skin is closer to the X-ray source than is the air kerma reference point, meaning that the reference air kerma displayed by the system will underestimate the patient’s skin dose. (b) Here the patient’s skin is further from the X-ray source. Consequently, the air kerma displayed by the system will overestimate the patient’s skin dose.

4

| 09.11.19 - 01:52

1.2 Case 2: Reference Air Kerma and Skin Dose distance. Consequently, entrance AK and actual patient skin dose may be larger or smaller than the reference AK. The patient’s position relative to the AK reference point is not the only factor that affects the PSD. If, for example, the C-arm is rotated during the fluoroscopy exam, the dose will be “spread out” over different areas of the patient’s skin. Similarly, if the table or C-arm is translated so that the imaging field of view moves from the groin to the chest, the total AK displayed by the system will include radiation delivered to different portions of the skin. While any detailed PSD estimate should be performed by a qualified medical physicist, understanding the relationship between the displayed AK and patient skin dose and recognizing when these values are likely to

diverge will allow the fluoroscopy operator to manage radiation dose during a procedure.

1.2.4 Resolution If the AK reference point is at 65 cm from the focal spot, then the AK at 53 cm from the focal spot is (65/53)2 or 1.5. This means that if the displayed AK is 5100 mGy, then the patient entrance AK will be approximately 7700 mGy. Other factors, such as attenuation of the X-ray beam by the table, backscatter generated by the patient, and a conversion of entrance skin exposure to absorbed skin dose must be considered to determine the PSD. However, entrance AK can be used as a reasonable surrogate for PSD when considering radiation management during a fluoroscopy procedure.

5

| 09.11.19 - 01:52

Fluoroscopy

1.3 Case 3: Collimation 1.3.1 Background ●

Patient underwent a fluoroscopy-guided exchange of a retrograde left nephroureteral tube.

●

Collimated (right)

Collimated field size

21.9 × 28.4 cm (620 cm2)

16.5 × 19.9 cm (328 cm2)

the procedure with different amounts of physical

DAP (mGycm2)

446

278

collimation.

AK (mGy)

1.3

1.3

Collimation was used to reduce the field of view (from 21.9 × 28.4 cm to 16.5 × 19.9 cm; ▶ Table 1.2).

●

Uncollimated (left)

Two digital spot images were acquired during

1.3.2 Findings ●

Table 1.2 The collimated field size and dose metrics for the uncollimated and collimated images shown in ▶ Fig. 1.2

Dose area product (DAP) was reduced by 47%.

1.3.3 Discussion This case illustrates that collimation (also sometimes referred to as “coning in”) can reduce radiation exposure to both patients and staff and improve image quality. Collimation uses lead shutters inside the X-ray tube housing to reduce the imaging field of view. This enables one to image (and irradiate) only the patient anatomy essential for performing the procedure.

Changes in collimation will not substantially affect AK. However, collimation will affect DAP (sometimes called Kerma Area Product, or KAP) which is another dose metric that is commonly displayed on fluoroscopy systems. DAP is the product of AK and field size, so even if AK remains constant, DAP will change proportionally to the change in X-ray field size. For procedures using single fields (i.e., the position of the X-ray tube remains constant relative to the patient throughout the exam), this will not reduce the PSD substantially but will reduce the effective dose to the patient. For procedures with multiple fields, proper collimation can reduce the likelihood of having overlapping radiation fields and potentially reduce PSD. In addition, collimation can greatly reduce staff exposure since a smaller imaging field will produce less scatter radiation.

Fig. 1.2 Digital spot images acquired without collimation (a) and with collimation (b).

6

| 09.11.19 - 01:52

1.3 Case 3: Collimation Finally, proper collimation has the additional benefit of enhanced image quality. Collimated fields can be used to exclude very high- or low-attenuating tissues, reducing the variation in signal intensity across the field of view, which may improve image contrast. Moreover, since proper collimation reduces the amount of scatter created by patient tissue, images will appear with improved contrast (Table 1.2).

1.3.4 Resolution In the example shown in ▶ Fig. 1.2, the field of view was reduced from 21.9 × 28.4 cm (620 cm2) in ▶ Fig. 1.2a to 16.5 × 19.9 cm (329 cm2) in ▶ Fig. 1.2b; a reduction of 47%. DAP also decreased 47%, i.e., from 807 mGycm2 (▶ Fig. 1.2a) to 428 mGycm2 (▶ Fig. 1.2a). AK was the same for both images (1.3 mGy).

7

| 09.11.19 - 01:52

Fluoroscopy

1.4 Case 4: Anti-scatter Grids 1.4.1 Background ●

A 5-year-old patient undergoes a fluoroscopyguided cardiac procedure.

●

The patient had a similar procedure 6 months prior where the total AK was 450 mGy.

●

With a similar amount of fluoroscopy time and the same number of exposures, the AK for the current procedure was 1200 mGy.

1.4.2 Findings ●

The same fluoroscopy system and imaging protocol were used for both procedures.

●

The images for the prior procedure (▶ Fig. 1.3a) have poorer image quality than the images from the later procedure (▶ Fig. 1.3b).

1.4.3 Discussion X-ray photons that reach the image detector are either primary or scattered photons. Primary photons travel along a straight path through the patient until they are absorbed by the detector. Scattered photons are those that have interacted with the patient’s tissue (or other objects) and have changed direction from their original path.

Image quality is greatly influenced by the scatter-to-primary ratio (SPR) or the ratio of the amount of energy deposited in the detector by scattered versus primary photons. If a fluoroscopy system has an SPR of 3, this indicates that 75% of the energy deposited in the detector is from scattered photons while the remaining 25% is from primary photons. As the SPR increases, low-contrast objects are more difficult to distinguish. Image quality can be improved by using an anti-scatter grid. The grids are made up of a series of septa made of lead or other X-ray-attenuating material, and these septa preferentially absorb scattered photons while allowing more of the primary photons to pass through. The grid is placed just in front of the image receptor, decreasing the SPR, and improving image quality. Since the grid also absorbs some of the primary photons, the X-ray system must compensate by producing more photons (i.e., increasing mAs), which increases patient dose. A main determinant of how much scatter is produced in a specific patient is patient habitus. As the X-ray beam passes through additional tissue, the number of scatter events increases. In other words, thicker anatomy and larger patients will produce more scatter than smaller patients. Consequently, the anti-scatter grid has a substantial effect on image quality when imaging adult patients, but it has little effect when imaging extremities or small patients.

Fig. 1.3 Fluoroscopy images of an anthropomorphic chest phantom without (a) and with (b) an anti-scatter grid.

8

| 09.11.19 - 01:52

1.4 Case 4: Anti-scatter Grids

1.4.4 Resolution ▶ Fig. 1.3 shows an image of an anthropomorphic chest phantom. The image on the left (▶ Fig. 1.3a) was obtained without an anti-scatter grid whereas the image on the right (▶ Fig. 1.3b) was obtained with the grid in place. The effect on image quality is apparent by looking at the increase in detail visible in the image on the right (obtained using the grid.) However, for some clinical indications the image quality in ▶ Fig. 1.3 may be sufficient. The acquisition parameters for the two images are also shown. The grid attenuates both primary and

scatter photons, causing the ABC system to increase both kV and mA. As a result, the AKR increased by a factor of almost 3. Removing the anti-scatter grid is a common way to reduce patient dose in fluoroscopy exams of pediatric patients, where image quality is not substantially affected. However, in most adult (or adult-sized) patients, the grid is required to achieve adequate image quality. It should be noted that while many fixed fluoroscopy systems have removable grids, most mobile C-arms have fixed grids that cannot be removed.

9

| 09.11.19 - 01:52

Fluoroscopy

1.5 Case 5: Patient Shielding 1.5.1 Background ●

A patient who is 30 weeks pregnant undergoes a fluoroscopy-guided intervention to treat iliofemoral deep venous thrombosis.

● ●

The patient is positioned prone on the table. A lead apron is placed under the patient’s abdomen and pelvis (between the patient and X-ray tube).

1.5.2 Findings ●

The lead apron covers most of the imaging field of view.

●

Image quality is diminished as contrast is reduced for important anatomical structures.

●

The fluoroscopy system’s ABC algorithm increases tube output to compensate for additional attenuation caused by the presence of the lead.

1.5.3 Discussion Shielding materials (typically lead aprons) are sometimes placed on or around patients with the goal of reducing radiation exposure to staff. This technique is also used with pregnant patients to reduce fetal exposure to radiation. However, placing lead under the patient may adversely

affects both image quality and dose. First, as shown in ▶ Fig. 1.4, the presence of the lead apron substantially reduces contrast throughout the image. Second, since lead is highly attenuating, it significantly affects the function of the ABC algorithm increasing the X-ray tube output and radiation dose to the patient, fluoroscopy operator, and other staff present during the case.2 Fetal radiation exposure during an exam of the mother varies substantially based on whether the fetus is in or outside the imaging field of view. If the fetus is in the imaging field of view, it is exposed to the primary X-ray beam and radiation dose could exceed 100 mGy, above which there is an increased the risk of congenital malformation, stillbirth, miscarriage, or mental disability.3 If the fetus is outside of the imaging field of view, the majority of radiation exposure is from scatter generated within the mother and the fetal dose is typically below 1 mGy.4 At such low dose levels, there is no demonstrated increased risk to the fetus. For a more detailed discussion of management of pregnant patient during fluoroscopy procedures, we refer the reader to multisocietal guideline published by Dauer et al.4 The use of patient shielding outside of the imaging field of view has been advocated as a way to reduce radiation dose to staff. However, it is important to keep in mind that the vast majority of operator exposure comes from scatter created by interactions of the primary beam with patient tissue. In other words, the radiation to which an

Fig. 1.4 A digital spot image of a pregnant patient who had lead apron placed between her skin and the X-ray tube in an attempt to provide additional protection to the fetus.

10

| 09.11.19 - 01:52

1.5 Case 5: Patient Shielding operator (and other staff) is exposed originates from the patient. Consequently, shielding placed on the patient can only provide a protective benefit when an operator is standing in a few specific locations. In addition, it has been demonstrated that even under these conditions, the protective benefit is negligible compared to that of the operator’s own protective garments (i.e., lead apron and glasses).5 Finally, due to the dynamic nature of fluoroscopically guided interventions, there is a risk that shielding initially positioned outside of the primary beam might end up in the imaging field of

view. This carries the same risks as shielding placed directly in the beam for patient protection, i.e., reduced image quality and increased radiation exposure to both the patient and staff.

1.5.4 Resolution In summary, shielding materials should not be placed on a patient. The benefit to the patient or operator is negligible, but the risk of compromising image quality is substantial. Furthermore, using patient shielding may increase radiation dose to everyone involved.

11

| 09.11.19 - 01:52

Fluoroscopy

1.6 Case 6: CT Fluoroscopy 1.6.1 Background ●

An obese patient with diabetes and a history of colon cancer presents with an enlarging right lower lobe nodule with metabolic activity on a recent positron emission tomography (PET)/ computed tomography (CT) scan.

●

The patient undergoes a CT-guided right lung mass biopsy (▶ Fig. 1.5).

●

A total of 30 scans are performed. The total CT dose index (CTDIvol) for the procedure is 1200 mGy. The dose length product (DLP) is 12,600 mGy-cm.

1.6.2 Findings The patient was referred to radiation safety for an estimate of PSD to determine if follow-up evaluation for radiation-induced skin injury was appropriate.

1.6.3 Discussion While traditional fluoroscopy is based on planar imaging techniques, CT fluoroscopy can be used to create “real-time” cross-sectional images to guide interventional procedures. Studies have demonstrated that CT-guided biopsy is a low-risk method of obtaining tissue samples of mediastinal masses.6 Dose metrics used in CT differ from those used in fluoroscopy. A basic understanding of CT dose metrics is a fundamental component of dose management in CT-guided interventional procedures. The radiation output of a CT scanner is commonly characterized by the CTDIvol and DLP. CTDIvol is an estimate of dose, in mGy, to a standard acrylic phantom. DLP is the product of CTDIvol and the scan length and is reported in units of mGy-cm. While these values can be used to estimate patient effective dose, neither corresponds well with skin dose, making the assessment of risk for radiation-induced skin injuries difficult. Studies have shown that skin dose can range from approximately 49 to 65% of the scanner-reported CTDIvol for scans performed without table movement (as is often the scenario during CT-guided interventional procedures).7 For a CT scanner, the X-ray tube cannot be moved closer or farther from the patient, meaning the

12

Fig. 1.5 A CT image obtained during the biopsy procedure. The arrow indicates where the biopsy needle is visible on the image.

X-ray tube will be closer to the patient’s skin if the patient is obese, compared with the SSD for a smaller patient. Consequently, if a patient is obese, the ratio of PSD to CTDIvol may be even higher than it would be for a smaller patient.

1.6.4 Resolution In the case presented, the total CTDIvol is 1200 mGy and the DLP is 12,600 mGy-cm. These dose metrics are considerably higher than those from a typical diagnostic CT scan of the adult abdomen, which has a CTDIvol of approximately 20 mGy and a DLP of approximately 1000 mGy-cm.8 Using the relationship cited above, where the PSD is approximately 65% of the reported CTDIvol, PSD can be conservatively approximated as 780 mGy. This is well below the point at which radiation-induced skin damage is expected.9 Of note, the patient in the case presented here has diabetes, which, due to the associated decrease in tissue perfusion, may increase skin radiosensitivity. For a discussion of other factors that can increase radiosensitivity, please see the article by Balter et al.9 DLP is commonly used to calculate effective dose.10 However, effective dose is not correlated with PSD. While CTDIvol is a better indicator of PSD, it suffers from some of the same limitations as cumulative AK reported on a fluoroscopy system, i.e., neither cumulative AK nor CTDIvol give any indication of how the radiation exposure was distributed across the patient’s skin. Although one can generally assume that the radiation from a

| 09.11.19 - 01:52

1.6 Case 6: CT Fluoroscopy CT-guided interventional procedure is evenly distributed around the circumference of the patient, when evaluating PSD, one should consider whether all imaging was performed over a single “slice” along the z-axis of the patient, or if the patient was moved superiorly or inferiorly during the procedure. Another consideration when performing CT-guided procedures is staff radiation exposure. Similar to traditional fluoroscopy, staff will only be exposed to the primary radiation beam if their body enters the imaging field of view. On the other

hand, staff radiation exposure usually comes from scatter created within the patient. Consequently, when possible, staff should stand to the side of the CT gantry, where the gantry provides a shielding barrier between the patient and the staff member. Alternatively, the operator can leave the room and perform a single-slice CT scan. This can result in a slightly higher radiation dose to the patient, but the cumulative occupational dose savings to an operator may be significant while the small increase in radiation exposure may not pose any additional risk to the patient.

13

| 09.11.19 - 01:52

Fluoroscopy

1.7 Case 7: Digital Subtraction Angiography and Motion Artifacts 1.7.1 Background ●

A patient undergoes a preoperative embolization of renal cell carcinoma metastasis in the left hand.

●

Digital subtraction angiography (DSA) is used to demonstrate reduced blood flow to the tumor following embolization.

1.7.2 Findings ●

During the DSA run, the patient repositions her hand, leading to significant artifacts.

●

DSA was repeated to obtain adequate image quality.

1.7.3 Discussion DSA is an imaging technique used to visualize blood vessels while minimizing the appearance of bone and soft tissues. It is performed by acquiring a “mask” image of the anatomy of interest before the administration of a contrast medium (typically iodine-based), followed by a series of postcontrast images. The mask image is subtracted from each image in the postcontrast series, reducing the visibility of overlapping tissues. When properly utilized, the resulting images provide a map of the vasculature of interest and have several uses in identifying stenosis, embolism, and aneurysm. Proper application of this technique requires that the anatomy being imaged is stationary in the imaging field of view. ▶ Fig. 1.6a shows the patient’s hand at the beginning of the DSA run before contrast has been administered. The hand Fig. 1.6 Individual frames from the beginning (a), middle (b), and end (c) of a DSA run where patient motion caused significant artifacts are shown. (d) The ability to visualize the vasculature of the hand was compromised leading to a repeated acquisition.

14

| 09.11.19 - 01:52

1.7 Case 7: Digital Subtraction Angiography and Motion Artifacts is well aligned with the previously acquired mask image. Consequently, the subtraction image masks the appearance of soft tissue and bone. ▶ Fig. 1.6b shows an image near the beginning of contrast administration. The radial artery is easily visualized as a result of contrast injection. However, as contrast is administered, the patient’s hand begins to move. This is most readily observed at the fingers where the misalignment between the current acquisition and the mask image is greatest. Finally, ▶ Fig. 1.6c shows an image that occurs near the end of the DSA run where the hand is now substantially mispositioned relative to the mask image. At this point in the run, the vasculature of the hand should be clearly visible but artifacts obscure the anatomy of interest. The initial position of the hand during the acquisition of the mask image can be readily identified in ▶ Fig. 1.6c by the lighter, almost white fingers and outer portion of the hand. Since the position of the hand has changed, the mask image is being subtracted from an area that no longer contains the anatomy. Consequently, this portion of the image is even brighter than the background. As a corollary, a darker image of the thumb and fingers is apparent, indicating the final position of the hand.

1.7.4 Resolution The example in ▶ Fig. 1.6 highlights the effects of motion in an exam of the hand. Extremities may be more prone to these artifacts since there is higher possibility of movement. However, any changes in patient position between acquisition of the mask image and the subsequent images will result in these effects. This can be caused by the operator moving the patient, table, or X-ray tube, or by involuntary patient motion (cardiac and respiratory). It should be noted that, in this case, the DSA run was repeated with reduced patient motion and image quality was greatly improved (▶ Fig. 1.6d). While extremities are not typically exposed to large amounts of radiation during fluoroscopy exams, procedures that are complex or require imaging through larger portions of the body require higher doses and have the potential to result in radiation injuries. These risks are more likely during DSA acquisitions compared to conventional fluoroscopy since the dose rate during DSA is substantially higher (even 20 times higher). Good practice is to limit the use of DSA to portions of the exam where it provides a clear clinical benefit over fluoroscopy and avoid repeating DSA acquisitions when possible.

15

| 09.11.19 - 01:52

Fluoroscopy

1.8 Case 8: Fluoroscopy Modes and Dose 1.8.1 Background ●

A patient with unstable angina received a cardiac catheterization procedure under fluoroscopic guidance.

●

Standard fluoroscopy, digital spot images, and cine mode fluoroscopy were used.

1.8.2 Findings The reference AKR during cine angiography was 28.5 times higher than during standard fluoroscopy.

1.8.3 Discussion Modern fluoroscopic systems have multiple operating modes, including standard fluoroscopy, highdose-rate (HDR) fluoroscopy, digital spot imaging, and cine (sometimes referred to as cinefluorography). The selection of an imaging mode during a procedure requires consideration of the image quality needs of the individual case as well as the relative radiation exposure rate for each mode. For all fluoroscopy systems, the maximum dose rate that can be physically produced by the system is limited by regulation for standard fluoroscopy and HDR modes.11 The location along the X-ray beam at which this is measured varies for different types of fluoroscopy systems. (For under-table systems, the maximum dose rate is measured at 1 cm above the patient table, while for C-arm type systems used in interventional radiology and cardiac catheterization labs, the maximum dose rate is measured at 30 cm from the image receptor.) When used in standard fluoroscopy mode, the maximum allowable dose rate is 88 mGy/minute.11 In HDR mode the limit doubles to 176 mGy/minute. While standard fluoroscopy produces adequate image quality for many patients, HDR may be required to maintain image quality in larger, more attenuating patients. In digital spot mode, the fluoroscopy system functions as a general radiography device producing a single image. The technique (kV, mAs) used to produce a digital spot image is generally

16

higher than that used for an individual fluoroscopy frame producing a higher-quality image but with an associated increase in dose. While the dose per image is substantially higher in digital spot mode, the total exposure to the patient may be less than an individual fluoroscopy run which may contain dozens of images. Finally, cine mode creates an “X-ray movie” similar to fluoroscopy but the image quality of each individual frame is roughly equivalent to that of a digital spot image. This improved image quality for the entire cine run comes at the expense of greatly increased dose rates. It should be noted that the dose rate in cine mode is not limited by regulation. This case illustrates the difference in dose between standard fluoroscopy and cine mode fluoroscopy. ▶ Fig. 1.7 shows two single-frame images of the same anatomic location. Collimation, X-ray tube and detector positions, and frame rate remained the same for both images. ▶ Fig. 1.7a was acquired as part of a fluoroscopy run, while ▶ Fig. 1.7b was part of cine run. The cine image (▶ Fig. 1.7b) clearly provides better image contrast as well as improved detail of high spatial frequency structures (edges). During the fluoroscopy run, the reference AKR was 20 mGy/minute. In contrast, the reference AKR during the cine run was 570 mGy/minute (more than 28 times the AKR during standard fluoroscopy).

1.8.4 Resolution It is important for fluoroscopy operators to carefully consider whether the imaging task requires the quality provided by cine and HDR fluoroscopy modes. Commonly, positioning of the imaging field of view relative to the patient can be performed using standard fluoroscopy with higher-dose-rate modes reserved for the anatomy of interest during select portions of the procedures. Positioning can also be aided by use of virtual collimators and positioning tools. Similarly, digital spot imaging yields a higher dose per image than fluoroscopy. All modern fluoroscopy devices are equipped with a “last image hold” function that presents the final image from any fluoroscopy run to the operator, providing a lower quality (and lower dose) static image. Digital

| 09.11.19 - 01:52

1.8 Case 8: Fluoroscopy Modes and Dose

Fig. 1.7 (a) A single frame from a fluoroscopy run performed during a cardiac catheterization procedure. (b) A cine image from the same procedure using the same positioning and collimation. While image quality is improved on the cine image, it comes at the expense of a substantial increase in dose rate. This image is similar to what would be seen from a digital spot.

spot images may then be reserved for situations where the improved image quality is necessary. As is the case with other dose reduction practices, imaging modes that result in lower patient dose also reduce staff dose. It should be noted that while the Food and Drug Administration (FDA) places limits on the maximum AKR that a system can produce while operating in fluoroscopy mode,11 no such limit exists when the system is used in other modes (cine, DSA, or single-shot acquisitions). While the

AKR in standard fluoroscopy mode was well below the regulated maximum of 88 mGy/minute (176 mGy/minute when operated in high-doserate fluoroscopy mode), the dose rate in cine mode was considerably higher. Interestingly, the reference point at which the maximum AK is regulated by the FDA is often different from the location of where the displayed reference AK is calculated. Consequently, the maximum displayed AKR may be higher than the regulated maximum AKR.

17

| 09.11.19 - 01:52

Fluoroscopy

1.9 Case 9: Equalization Filters

●

space,” where there is no anatomy to attenuate

1.9.1 Background ●

A patient with a history of primary sclerosing

The left portion of the image contains “open the X-ray beam.

●

Contrast is poor inferior to the lungs.

cholangitis, coagulopathy, and chronic portal vein thrombosis undergoes an

1.9.3 Discussion

intrahepatic portosystemic shunt creation

Section 1.4, Case 3: Collimation, discussed the use of collimators to limit the X-ray field to the anatomy of clinical interest. These collimators are made of lead and are thick enough to fully attenuate the X-ray beam. In addition to collimators, many fluoroscopy systems have equalization filters. These filters are sometimes known as wedges or contour filters. As the name implies, wedge filters have varying thickness and are typically made of lead-impregnated acrylic sheets. They are less absorptive than collimators and provide more attenuation through the “heel” portion of the wedge than through the “toe.” The filters can be moved in and out of the imaging field view over structures that are less attenuating. To understand the effect that equalization filters have on image quality, one must consider how digital images are displayed. The ultimate purpose of these physical filters is to equalize the signal intensity across the image. Since a finite number of grayscale levels are available to display a digital image, having a wide range of signal intensities across the field of view results in having fewer grayscale values available to display less-attenuating anatomy, thus degrading image contrast in these areas. Proper use of wedge filters can help to maintain image contrast for exams where the attenuation properties of objects in the imaging field of view vary substantially. The case presented here highlights the effect of not using equalization filters. The lungs are less attenuating relative to other anatomic structures in the image, that is, the diaphragm and vertebral bodies. Without equalization filters, the lungs appear extremely bright and washed out.

procedure. ●

Much of the procedure requires imaging over an area that includes the lungs, liver, and bony structures of the vertebral bodies.

1.9.2 Findings ●

The lung appears hyperintense, obscuring any anatomical contrast in the lungs (▶ Fig. 1.8).

Fig. 1.8 The lung in this image appears extremely bright. The image also contains space to the patient’s left which also appears hyperintense. The image contrast in other regions of the image is poor.

18

1.9.4 Resolution Proper use of wedge filters can improve image quality, especially in cardiac and thoracic imaging,

| 09.11.19 - 01:52

1.9 Case 9: Equalization Filters where an image often contains portions of the heart, lungs, and bone, all of which have vastly different X-ray attenuation properties. ▶ Fig. 1.9 provides an example of the benefits of using wedge filters to equalize the signal intensity across the image. The wedge filters are placed over portions of the right and left lungs, improving image quality across the entire image. An added benefit of using equalization filters is a reduction in radiation dose to areas covered by the filters. Another way to achieve equalization of attenuation properties across the image is to place a bolus material next to the patient. In the case shown in ▶ Fig. 1.8, a bolus placed next to the patient’s chest would decrease the brightness on the left side of the image. However, bolus materials increase scatter, degrading image quality and increasing patient and operator dose. For this reason, wedge filters are typically preferred to achieve image equalization.12

Fig. 1.9 An image showing wedge filters positioned over the left and right portions of the image. By partially covering the radiotransparent lungs, image quality is improved across the entire field of view.

19

| 09.11.19 - 01:52

Fluoroscopy

1.10 Case 10: Cone Beam Computed Tomography 1.10.1 Background ●

A patient has a history of massive hemoptysis and blood visualized in the right bronchus during bronchoscopy.

●

An embolization procedure is performed, including a right bronchial artery angiogram.

●

Cone beam computed tomography (CBCT) imaging is used to provide axial and coronal images of the liver and localize the tumor for therapy.

1.10.2 Findings Streaking and motion artifacts are clearly visible on both the axial (▶ Fig. 1.10a) and coronal (▶ Fig. 1.10b) images reconstructed from the CBCT acquisition.

1.10.3 Discussion CBCT is used in the interventional environment to produce cross-sectional and 3D reconstructions. This intraprocedural imaging technique is used to visualize anatomy and guide intervention. In some procedures, such as transarterial chemoembolization (TACE) for treatment of liver cancer, CBCT

has been demonstrated to be superior to DSA for visualization of liver tumors.13 CBCT images can also be compared with images obtained prior to the procedure using either MDCT or magnetic resonance imaging. Similar to standard multidetector CT (MDCT), CBCT utilizes projections taken at multiple angles around the anatomy of interest to reconstruct axial, sagittal, and coronal slices. As with other tomographic techniques, anatomic visualization is improved by removing overlapping tissue. Since the principles of CBCT are similar to those for MDCT, CBCT is susceptible to similar image artifacts. However, the clinical implementation of CBCT may result in unique manifestations of these artifacts. ▶ Fig. 1.10 shows axial (▶ Fig. 1.10a) and coronal (▶ Fig. 1.10b) images using CBCT during a liver embolization procedure. In the axial image, streaking artifacts result from interactions between the primary X-ray beam and the metal catheter placed as part of the procedure. Incident X-rays are highly absorbed by the catheter, resulting in fewer X-ray photons reaching the detector. The star pattern is due to this effect occurring in multiple X-ray projections. In addition, in the coronal image, a possible beam hardening artifact is shown inferior to a region of hyperintensity. The images in ▶ Fig. 1.10 also suffer from blurriness, especially visible in the lungs which is caused

Fig. 1.10 Axial (a) and coronal (b) images using cone beam computed tomography are shown. Both have noticeable streaking and motion artifacts.

20

| 09.11.19 - 01:52

1.10 Case 10: Cone Beam Computed Tomography by breathing motion. While motion artifacts also occur during MDCT, the acquisition times of CBCT tend to be longer, leaving more time for motion to occur. (On modern systems, CBCT acquisition times are generally around 6–10 seconds. Rotation times in MDCT are often below 1 second.)

1.10.4 Resolution Adequate patient breath holds, starting a couple of seconds prior to initiation of imaging, are essential during CBCT. ▶ Fig. 1.11 shows axial and coronal images obtained using a patient breath hold. Compared with the images shown in ▶ Fig. 1.10, motion artifacts are markedly reduced. As with MDCT, the ability to achieve an adequate breath hold may be limited by the patient and procedure being performed. An additional consideration is that patients are more likely to be under some

level of sedation during CBCT, so the amount of sedation should be considered in evaluating breath hold options and limitations. Compared with MDCT, CBCT suffers from low contrast-to-noise ratio. However, this limitation is not generally clinically significant because the use of iodinated contrast greatly increases the visibility of the relevant anatomy. CBCT tends to have better inherent spatial resolution, since the pixel pitch of the flat panel detectors on interventional angiography systems is smaller than the detectors used in MDCT. (A modern flat panel detector has a pixel pitch of approximately 0.15 mm; a MDCT detector is often around 0.6 mm.) Finally, studies have found that CBCT results in lower patient skin dose compared with DSA.13 CBCT also often results in shorter overall imaging times, further reducing radiation dose to both patients and staff.

Fig. 1.11 Axial (a) and coronal (b) images using cone beam computed tomography with breath hold are shown. These images have fewer motion artifacts than those shown in ▶ Fig. 1.10.

21

| 09.11.19 - 01:52

Fluoroscopy

1.11 Review Questions 1.11.1 Case 1: SID, ABC, and Radiation Output 1. During a fluoroscopy-guided interventional procedure, the operator increases the SID from 90 to 120 cm without changing SSD. What effect will this change have on entrance skin dose? a) Entrance skin dose will increase by greater than 50%. b) Entrance skin dose will increase by less than 50%. c) Entrance skin dose will be unaffected. d) Entrance skin dose will be reduced by less

4. If the patient table is raised from a height of 50 cm above the X-ray source to 60 cm above the X-ray source, and all other imaging parameters remain the same, how much will the displayed AK change? a) The displayed AK will be 1.4 times the AK displayed at 50 cm. b) The displayed AK will be 1.2 times the AK displayed at 50 cm. c) The displayed AK will remain the same. d) The displayed AK will be 0.8 times the AK displayed at 50 cm. e) The displayed AK will be 0.7 times the AK displayed at 50 cm.

than 50%. e) Entrance skin dose will be reduced by greater than 50%.

1.11.3 Case 3: Collimation 5. For a procedure performed using a single field

2. When moving from imaging a patient’s torso

(i.e., without any tube rotation), collimating the

to imaging the abdomen, the ABC algorithm

imaging field of view from 600 to 400 cm2 will

used by the fluoroscopy system will likely

have what effect on PSD?

compensate by doing which of the following?

a) PSD will decrease approximately by 33%.

a) Lowering kV

b) PSD will not change.

b) Increasing mA

c) PSD will increase approximately by 33%.

c) Adding filtration d) Decreasing magnification

6. For a procedure performed using a single field, collimating the imaging field of view from 600

1.11.2 Case 2: Reference Air Kerma and Skin Dose 3. During a pain management procedure using a C-arm with a fixed SID, the operator raises

to 400 cm2 will have what effect on scatter dose to the operator? a) Operator dose will decrease by 33% b) Operator dose will not change. c) Operator dose will increase by 33%.

the patient table such that the SSD changes from 50 to 75 cm. What effect will this change have on entrance skin dose? a) Entrance skin dose will increase by greater than 50%. b) Entrance skin dose will increase by less than 50%. c) Entrance skin dose will be unaffected. d) Entrance skin dose will be reduced by less than 50%. e) Entrance skin dose will be reduced by greater than 50%.

22

1.11.4 Case 4: Anti-scatter Grids 7. For what clinical exam would it be appropriate to remove the anti-scatter grid? a) Sacroiliac joint injection b) Endoscopic retrograde cholangiopancreatogram c) Inferior vena cava filter placement d) Upper extremity arteriovenous fistulagram

| 09.11.19 - 01:52

1.11 Review Questions 8. What is the primary consequence of removing

c) 3000 mGy

the anti-scatter grid for an adult abdomen

d) 6000 mGy

protocol? a) Decreased image quality b) Increased patient skin dose c) Decreased visualization of iodine contrast d) Increased DAP

1.11.7 Case 7: Digital Subtraction Angiography and Motion Artifacts 13. In DSA, why are blood vessels readily visible while the appearance of other tissues is

1.11.5 Case 5: Patient Shielding 9. What is one of the most effective ways to decrease radiation exposure to a fluoroscopy operator? a) Wear appropriate lead garments (e.g., apron, glasses). b) Position a lead drape on the patient. c) Increase the SID.

minimized? a) The kV of the mask and second acquisitions are adjusted to specifically remove certain tissues types following subtraction. b) The addition of contrast agents to the blood vessels between the mask and the second acquisition highlights the vessels. c) The attenuation properties of blood vessels are significantly different than those of other soft tissue, making them more visible.

10. When performing a cardiac fluoroscopy-guided procedure on a pregnant patient, what is the

14. During DSA, an X-ray attenuating object is

primary source of radiation exposure to the

removed from the imaging field of view after

fetus?

the mask image is acquired but before the

a) Primary X-ray beam

second acquisition. What effect will it have on

b) Internal scatter

the images resulting from subtraction?

c) Backscatter

a) The object will leave a dark shadow relative to background.

1.11.6 Case 6: CT Fluoroscopy 11. What is the primary reason the DLP for a CT-guided fluoroscopy-guided procedure is typically higher than for a diagnostic CT scan? a) The total CTDIvol is typically higher for a CT fluoroscopy-guided procedure. b) The scan length is typically higher for a CT fluoroscopy-guided procedure. c) A higher kV is typically used for a CT fluoroscopy-guided procedure. d) A higher effective mAs is typically used for a CT fluoroscopy-guided procedure.

b) The object will have no effect on the subtraction image since it is removed before the second acquisition. c) The object will leave a light shadow relative to background.

1.11.8 Case 8: Fluoroscopy Modes and Dose 15. The patient dose from a digital spot image is roughly equivalent to which of the following images? a) The dose from a single fluoroscopy frame b) The dose from an HDR fluoroscopy frame

12. Radiation-induced skin injuries may occur at PSDs as low as 2000 mGy. For an average-sized

c) The dose from a single cine frame 16. For what imaging mode is the maximum AKR

adult patient, what CTDIvol would be associated

limited to 176 mGy/minute?

with a PSD of 2000 mGy?

a) Standard fluoroscopy

a) 1300 mGy

b) HDR fluoroscopy

b) 2000 mGy

c) Cine mode fluoroscopy

23

| 09.11.19 - 01:52

Fluoroscopy

1.11.9 Case 9: Equalization Filters

References

17. During a cardiac imaging procedure, the lungs

[1] AAPM Report 125. Functionality and operation of fluoroscopic automatic brightness control/automatic dose rate control logic in modern cardiovascular and interventional angiography systems (2012) [2] Marsh RM. What considerations should be made when performing fluoroscopy-guided interventions on pregnant patients? AJR Am J Roentgenol. 2017; 209(3):W195–W196 [3] International Commission on Radiation Protection. Biological effects after prenatal irradiation (embryo and fetus). ICRP Publication 90. Ann ICRP. 2003; 33:1–206 [4] Dauer LT, Thornton RH, Miller DL, et al. Society of Interventional Radiology Safety and Health Committee, Cardiovascular and Interventional Radiology Society of Europe Standards of Practice Committee. Radiation management for interventions using fluoroscopic or computed tomographic guidance during pregnancy: a joint guideline of the Society of Interventional Radiology and the Cardiovascular and Interventional Radiological Society of Europe with Endorsement by the Canadian Interventional Radiology Association. J Vasc Interv Radiol. 2012; 23(1):19–32 [5] Smith JR, Marsh RM, Silosky MS. Is lead shielding of patients necessary during fluoroscopic procedures? A study based on kyphoplasty. Skeletal Radiol. 2018; 47(1):37–43 [6] Petranovic M, Gilman MD, Muniappan A, et al. Diagnostic yield of CT-guided percutaneous transthoracic needle biopsy for diagnosis of anterior mediastinal masses. AJR Am J Roentgenol. 2015; 205(4):774–779 [7] Leng S, Christner JA, Carlson SK, et al. Radiation dose levels for interventional CT procedures. AJR Am J Roentgenol. 2011; 197(1):W97–103 [8] Kanal KM, Butler PF, Sengupta D, Bhargavan-Chatfield M, Coombs LP, Morin RL. U.S. diagnostic reference levels and achievable doses for 10 adult CT examinations. Radiology. 2017; 284(1):120–133 [9] Balter S, Hopewell JW, Miller DL, Wagner LK, Zelefsky MJ. Fluoroscopically guided interventional procedures: a review of radiation effects on patients’ skin and hair. Radiology. 2010; 254(2):326–341 [10] AAPM Report 96. The measurement, reporting, and management of radiation dose in CT (2008) [11] Performance standards for ionizing radiation emitting products, 21 C.F.R. §1020.32 (2018) [12] Schueler BA. The AAPM/RSNA physics tutorial for residents: general overview of fluoroscopic imaging. Radiographics. 2000; 20(4):1115–1126 [13] Tacher V, Radaelli A, Lin M, Geschwind JF. How I do it: conebeam CT during transarterial chemoembolization for liver cancer. Radiology. 2015; 274(2):320–334

appear so bright on the images that the cardiac structures are poorly visualized. What is the most appropriate method for improving image quality? a) Positioning a bolus material under the patient b) Using equalization filters over the lungs c) Decreasing the amount of copper filtration d) Increasing the collimated field of view 18. What is the primary benefit of using equalization filters? a) Improved image contrast b) Increased patient dose c) Improved spatial resolution d) Increased temporal resolution

1.11.10 Case 10: Cone Beam Computed Tomography 19. What is the primary advantage of using CBCT instead of DSA to view liver tumors during interventional procedures? a) Decreased patient skin dose b) Decreased motion artifacts c) Improved spatial resolution d) Improved visualization of lesions 20. What is the primary limitation of CBCT compared with MDCT? a) Poor spatial resolution b) Increased patient dose c) Limited imaging field of view d) Introduction of streaking artifacts

24

| 09.11.19 - 01:52

2 Mammography Ingrid S. Reiser

Introduction Mammography uses X-ray projection imaging to acquire images of the breast. Mammography has been the gold standard for breast cancer screening since the 1990s. In addition to breast cancer screening, mammography is used for diagnostic breast imaging. Conventional mammography produces 2D projection images of the breast. The complex anatomical background of structures can hinder tumor detectability. With the advent of digital full-field detectors for mammography, digital breast tomosynthesis (DBT) became feasible. DBT produces quasi-3D images of the breast that provide some depth resolution to overcome limitations of geometric superpositioning in conventional 2D projection images. DBT has helped reduce callback rates from screening mammography and potentially helps increase cancer detection rates. Mammography utilizes ionizing radiation to detect breast cancers. Since ionizing radiation also induces cancer (albeit at a much lower rate), it is important to limit and monitor the average dose to glandular breast tissues during mammography.

Common Image Quality Problems Mammography is a unique imaging modality in the sense that it is dedicated to and optimized for a single anatomy. The energy of the X-ray beam,

X-ray absorption, and resolution of the detector are optimized for the detection of low-contrast tumors and microcalcifications in the breast. X-ray energies impact radiographic contrast and patient dose, and are optimized to produce images with the best tumor signal-to-noise ratio at the lowest dose. Over- or underexposure of the mammogram due to incorrect X-ray technique is less common in digital detectors than film because of their wide dynamic range. However, there are a number of detector artifacts in digital mammography that can potentially mimic or obscure suspicious findings that must be recognized. In DBT, a quasi-3D image volume is synthesized from a series of lowdose projection images acquired with the X-ray tube moving across an arc of 15 to 50 degrees (depending on vendor implementation). The limited angle acquisition provides some depth resolution, but also causes artifacts in the tomosynthesis image volume. As in radiography, the most common problem affecting image quality is patient positioning and patient motion. Further, image quality can be affected by artifacts that mimic or obscure relevant anatomy. Speck-like artifacts can potentially mimic microcalcifications. There are multiple causes for these artifacts as described below. Artifacts caused by the image receptor include row dropout and electromagnetic interference (EMI). Mammography utilizes a grid in contact mode. Artifacts due to grid positioning, such as grid cutoff, do not occur because the grid and X-ray source are in a fixed geometry.

25

| 09.11.19 - 01:52

Mammography

2.1 Case 1: Magnification Imaging 2.1.1 Background Magnification views are acquired when visualization of fine detail is required. Magnification views are often used in the diagnosis of microcalcifications. ▶ Fig. 2.1 demonstrates this effect. A microcalcification cluster is imaged in magnification mode (▶ Fig. 2.1a), which is shown enlarged in ▶ Fig. 2.1b. ▶ Fig. 2.1c shows the microcalcification cluster imaged in contact mode. The resolution in ▶ Fig. 2.1c is markedly lower and as a result less detail is perceived.

2.1.2 Findings The images shown in ▶ Fig. 2.1a, b were acquired with the breast placed on the magnification stand at a height of 31 cm, resulting in a magnification factor of 1.8 for a source-to-detector distance of 70 cm. An increase in resolution is

gained from the magnification setup that includes use of a smaller focal spot compared to contact mode to minimize focal spot blurring.1

2.1.3 Discussion In magnification mode, the breast is positioned on a magnification stand and is thus located closer to the X-ray source (▶ Fig. 2.1d). As a result, structures within the breast are magnified by a factor M, depending on the distance between the X-ray focal spot and the breast on the magnification stand (dmag), and the distance between the X-ray focal spot and the detector (ddet). The magnification factor M is given by the ratio of these distances as M = ddet/dmag.

2.1.4 Resolution The height of the magnification stand determines the magnification factor. Most mammography systems allow for several heights of the magnification stand.

Fig. 2.1 Magnification views improve visualization of fine detail. Microcalcification cluster magnification imaging (a, b) and in contact imaging (c). The magnification factor in this example was 1.8. Geometry of magnification imaging: object magnification (d).

26

| 09.11.19 - 01:52

2.2 Case 2: Focal Spot Size Selection in Magnification Views

2.2 Case 2: Focal Spot Size Selection in Magnification Views 2.2.1 Background ●

Magnification views can help visualize small structures, such as microcalcification clusters, with greater detail. In contact imaging, when the breast is placed directly on the breast support, the system uses a large focal spot (typically 0.3 mm).

●

When the breast is positioned on the magnification stand and magnification mode is used, the system switches to a small focal spot (typically 0.1 mm).

2.2.2 Findings ▶ Fig. 2.2a, b shows a comparison of a test phantom imaged with a large (0.3 mm) and a small (0.1 mm)

focal spot. When the phantom is placed on the breast support (▶ Fig. 2.2a), the images acquired with a large or small focal spot exhibit similar sharpness. When the test phantom is placed on the magnification stand, the image acquired with the large focal spot is less sharp due to focal spot blur.

2.2.3 Discussion The focal spot of an X-ray tube is not a single point, but a small area from where X-rays are emitted. In mammography, the size of the focal spots are 0.3 mm (large) and 0.1 mm (small). The finite size of the focal spot results in blurring of the image.1 This is demonstrated in ▶ Fig. 2.2a, b and depends on the distance between the focal spot, the breast, and the detector. When the object (i.e., breast) is located in close contact with the detector and the focal spot is far away, focal spot blur is negligible for both the large and the small focal spot. Therefore, the images shown in ▶ Fig. 2.2a are equally sharp. In magnification

Fig. 2.2 (a) Phantom images acquired in contact mode with different focal spot sizes (FS). (b) Phantom images acquired in magnification mode with different focal spot sizes (FS). (c) Focal spot blur in contact mode and magnification mode.

27

| 09.11.19 - 01:52

Mammography mode, the breast is half-way between X-ray focal spot and the detector. In this configuration, the size of the projected focal spot is equal to the actual focal spot size and blurs the image accordingly. This is shown in ▶ Fig. 2.2b. The differences in blur become visible when focal spot blur is the dominant factor that limits image resolution. In contact imaging, shown in ▶ Fig. 2.2a, image sharpness is limited by detector resolution, and therefore no difference in image sharpness is observed when the focal spot sizes are changed. ▶ Fig. 2.2c shows how the distance of the object from the X-ray focal spot, dmag, affects focal spot

28

blur. Focal spot blur can be computed from the imaging distances as:
 ddet  dmag wblur ¼ w  dmag where w is the size of the focal spot, and wblur is the width of focal spot blur.

2.2.4 Resolution The small focal spot needs to be selected when performing imaging in magnification mode to optimize resolution of the system.

| 09.11.19 - 01:52

2.3 Case 3: X-ray Acquisition Technique Factors in Mammography

2.3 Case 3: X-ray Acquisition Technique Factors in Mammography 2.3.1 Background In mammography, technique factors, such as mAs, kVp, filtration, are chosen depending on compressed breast thickness and composition (fatty/heterogeneous/dense). Higher kV is chosen for larger breast thicknesses.

2.3.2 Findings ●

Mammograms of two different patients are shown in ▶ Fig. 2.3a, b. The compressed breast thicknesses are 35 and 82 mm. The patients were imaged on the same mammography system using automated exposure control.

●

X-ray technique factors (mAs, kV, filter) for the acquisition of these two images are shown in the figure, along with half-value layer (HVL) and average glandular dose (AGD).

●

The system selected higher mAs, kV, and a different filter to image the 82-mm breast, resulting in a greater HVL and AGD.

●

The ACR digital mammography accreditation phantom was imaged at equal average glandular dose but different kV and target/filtration settings, and image quality was compared (▶ Fig. 2.3c, d).

2.3.3 Discussion As the X-ray tube potential (i.e., kV) increases, subject contrast is reduced. This is demonstrated in ▶ Fig. 2.3c, where the contrast of all masses is greater for the 28 kV W/Rh image, compared to 37 kV W/Ag. As the X-ray tube potential (i.e., kV) increases, contrast-to-noise ratio reduces, as shown in ▶ Fig. 2.3d. The display window width is set to achieve equal displayed contrast of the largest mass, compared to the background. The 37-kV image is noisier, and it is more difficult to see the smaller masses and fibers. Technique factors are set so as to optimize the contrastto-noise ratio per average glandular dose. For Fig. 2.3 (a, b) Mammographic technique for different breast sizes. (c, d) Signal insert of the full-field digital mammography accreditation phantom imaged at equal average glandular dose (AGD), but different filtration and kV settings. In (c), the display window width is equal for both images. In (d), the display window is different for both kV settings and was chosen to produce equal contrast of the largest mass in both images.

29

| 09.11.19 - 01:52

Mammography the phantom images, the optimum X-ray technique is 28 kV compared to 37 kV, because it gives better image quality in terms of higher contrast and contrast-to-noise ratio. When imaging a thicker breast, generally a higher kV is used to achieve a more penetrating beam. This is manifested in the increased HVL of the X-ray beam. Improved image quality could also be achieved by increasing mAs to maintain subject contrast, but prolonged exposure times can lead to artifacts from patient motion (see Case 9, Patient Motion Causing Blurred Parenchymal

30

Structure in a Mammogram), and it can also potentially cause X-ray tube heat overloading. In addition, increasing mAs increases dose. Average glandular dose is linearly proportional to mAs, i. e., doubling mAs produces twice the AGD.

2.3.4 Resolution kV should be increased with breast thickness and breast density to produce a more penetrating X-ray beam that optimizes subject contrast.

| 09.11.19 - 01:52

2.4 Case 4

2.4 Case 4: Digital Breast Tomosynthesis: Artifacts due to High-Contrast Objects

seen as it fans out as the distance to the focus slice increases. The angle of the fan is that of the tomosynthesis scan.

2.4.1 Background

2.4.3 Discussion

Tomosynthesis imaging is a quasi-3D imaging modality of the breast. The X-ray source travels along an arc of 15 to 50 degrees, depending on vendor implementation, and a series of lowdose projections are acquired. The tomosynthesis images are reconstructed from these projections.

In tomosynthesis, depth resolution is achieved by blurring structures below and above the in-focus plane, and enhancing structures that are truely located at that depth. High-contrast structures often persist throughout all tomosynthesis images and present themselves as repeat “copies” of the object, as is the case for the microcalcification shown in ▶ Fig. 2.4a. Sometimes, a calcified vessel appears as ripples in slices above or below the slice of focus. Other high-contrast objects that can produce such artifacts include biopsy markers.

2.4.2 Findings ●

▶ Fig. 2.4a shows a large calcification as it appears in the focus plane, i.e., in the tomosynthesis image at the actual depth of the object. The insets show slices displaying the calcification at different depths. The calcification is clearly seen at other depths, but it becomes more distorted as the tomosynthesis image is further away from the in-focus depth.

●

▶ Fig. 2.4b shows a perpendicular slice through the breast volume. The calcification is clearly

2.4.4 Resolution The limited angle scan geometry of tomosynthesis causes artifacts from high-contrast objects, such as large calcifications. Depending on the image reconstruction and image processing used by the equipment manufacturer, conspicuity of these artifacts can vary. It is important to understand the origin of these artifacts.

Fig. 2.4 (a) Artifact due to a high-contrast calcification in tomosynthesis volume. Regions centered on the calcification are shown at different depths. Repeated ghosts of the calcification can be observed at depths far from its in-focus plane. (b) Perpendicular slice through the tomosynthesis volume at the level of the high-contrast calcification. The X-ray tube moves along the X-direction during the tomosynthesis scan.

31

| 09.11.19 - 01:52

Mammography

2.5 Case 5: Effect of Image Post-Processing on the Appearance of a Mammogram

in the image falls off toward the skin line because breast thickness decreases. ●

patient but acquired on full-field digital

2.5.1 Background ●

mammography (FFDM). The breast is visualized up to the skin line, which was achieved

Digital mammography uses image processing

through a uniformity correction. As a result,

to enhance contrast.2,3 ●

mammogram, which virtually eliminate the need to adjust display window width and level. ●

Mammograms obtained from different vendor systems can have a significantly different “look” due to proprietary choices in image processing and presentation.

2.5.2 Findings ●

structures close to the skin line are still clearly

Processed mammograms are overall more uniform in comparison to a screen-film

▶ Fig. 2.5b shows a mammogram of the same

visible. ●

▶ Fig. 2.5c shows a mammogram acquired on a FFDM unit of a different vendor. As in ▶ Fig. 2.5b, the breast is visualized up to the skin line. However, the image is processed to show greater contrast compared to that shown in ▶ Fig. 2.5b. ▶ Fig. 2.5d shows a synthetic 2D image generated from tomosynthesis volume images for the same patient as in ▶ Fig. 2.5c. The synthetic mammogram is similar but not equal to a conventional 2D

▶ Fig. 2.5a–d shows images from a patient

mammogram. In the example shown, the calci-

acquired on different vendor systems. ▶ Fig. 2.5a

fication exhibits dark overshoots above and

shows a screen-film mammogram. The intensity

below it, which is an artifact of tomosynthesis

Fig. 2.5 (a–d) Image processing affects the appearance of mammograms. A patient was imaged with different systems: screen-film mammography (a), full-field digital mammography (FFDM) vendor A (b), FFDM vendor B (c), 2D mammogram synthesized from tomosynthesis (d). Note the difference in the calcification (arrows) shown in the inset in (c) and (d). (Continued)

32

| 09.11.19 - 01:52

2.5 Case 5: Effect of Image Post-Processing on the Appearance of a Mammogram

Fig. 2.5 (Continued) (e) Fourier-domain filters can enhance large or small detail in an image.

Fig. 2.5 (Continued) (f–h) Original mammogram (f) processed with different unsharp masking parameters (g, h). The image uniformity is improved, which allows display of the image at a higher display contrast in (h), while maintaining visibility of the skin line.

33

| 09.11.19 - 01:52

Mammography that is not seen in the conventional 2D mammogram (▶ Fig. 2.5c).

2.5.3 Discussion In digital mammography as in digital radiography, the image consists of an array of numeric values. Image processing is achieved through mathematical operations on these numbers (i.e., pixel values). Two basic image processing operations, by use of the Fourier transform and by histogram processing, are demonstrated below. The Fourier transform is a basic element of image processing that takes advantage of the spatial frequency. Low spatial frequencies represent large structures and image contrast, while high spatial frequencies represent image detail such as fine structures and edges, and image noise. A lowpass filter selects large structures in an image, while a high-pass filter selects image detail and

34

enhances edges. The effect of these filters is demonstrated in ▶ Fig. 2.5e. In unsharp masking, a low-pass filtered copy of the original image is subtracted from the image, producing a sharper and more uniform image ▶ Fig. 2.5f-h. Digital mammograms are processed to achieve greater uniformity, eliminating the intensity drop-off near the skin line that is observed in screen-film mammograms. In addition, images might be processed to produce a greater displayed contrast.2

2.5.4 Resolution Image processing algorithms are proprietary and can differ between mammography equipment vendors. Differences in image appearance between vendors are more likely caused by different image post-processing algorithms than differences in acquisition technique or dose.

| 09.11.19 - 01:52

2.6 Case 6: Artifact due to Detector Row Dropout

2.6 Case 6: Artifact due to Detector Row Dropout

dropout forms a band corresponding to the line dropout visualized in the tomosynthesis images at different depths.

2.6.1 Background Detector artifacts can be caused by individual pixel dropout or by an entire detector row dropout.4,5 This case describes the appearance of detector row dropout in conventional 2D mammography and 3D tomosynthesis and associated synthetic 2D view.

2.6.2 Findings ●

In this example, dropout is observed for two detector rows. ▶ Fig. 2.6a shows the appearance of the artifact in the conventional mammogram. As expected, two lines are seen.

●

▶ Fig. 2.6b shows the appearance of the artifact

2.6.3 Discussion The appearance of the dropout artifact differs in the conventional 2D mammogram and the simulated 2D image. This underlines the fact that the synthetic 2D image is created from the 3D tomosynthesis volume. Depending on vendor implementation, the synthetic 2D image may not be intended to mimic a 2D mammogram, but instead enhances potentially suspicious features that are seen in the tomosynthesis image, such as edges. In this case, edge post-processing likely enhanced the line artifacts and caused the banding in the synthetic 2D image.

in 3D tomosynthesis. In this particular case, row dropout occurred at two locations in one single projection view. In the tomosynthesis images, two lines can be observed and the location of the lines changes in different depths. In the simulated 2D image, which incorporates information from the 3D volume, the row

2.6.4 Resolution The dropout signal observed in these images represents a failure in the readout hardware of the digital detector and could not be remediated by recalibrating the detector. In this case, the detector panel needed to be replaced.

Fig. 2.6 (a) Line artifacts due to detector row dropouts in a conventional mammogram. In this case, two lines are seen. (b) Line artifacts due to detector row dropout in tomosynthesis. In this example, row dropout occurred in a single projection view. In the tomosynthesis images, two lines are seen. The lines sweep across the tomosynthesis images while scrolling through different depths. In the corresponding synthetic 2D image, the lines appear as bands.

35

| 09.11.19 - 01:52

Mammography

2.7 Case 7: Microcalcification-like Appearance Caused by a Detector Artifact 2.7.1 Background ●

An unexpected microcalcification-like object was observed during ACR phantom image quality review.

●

Prior patient images were reviewed. An artifact was found at the identical pixel locations.

2.7.2 Findings ●

The defect is seen in prior patient images at the same pixel location.

●

In the clinical example shown below, the artifact appears dark, while it appears white in the phantom image.

2.7.3 Discussion Small speck-like artifacts can be observed in mammograms. Their cause can be multiple (also see Case 8, Artifact due to Imperfection in Compression Paddle). An artifact in the detector manifests itself in the same location in different images. This artifact can be distinguished from dust or scratches by wiping all surfaces and rotating the test phantom. An artifact in the detector will always remain at the same location. Multiple quality control tests might reveal this artifact, including a weekly artifact test during which an image of a uniform phantom is inspected, as well as the phantom image quality test.

2.7.4 Resolution Detector artifacts can be resolved by updating the dead pixel map, which is used to mask out bad pixels. If there are too many bad pixels or the region of bad pixels is too large, the manufacturer might choose to replace the detector (▶ Fig. 2.7).

Fig. 2.7 Dead pixel artifact in the detector in (a) patient image and (b) ACR mammography accreditation phantom image. Artifacts in the detector occur at the same location in the patient image, and they might appear as dark or bright, or mixed (see insets).

36

| 09.11.19 - 01:52

2.8 Case 8: Artifact due to Imperfection in Compression Paddle

2.8 Case 8: Artifact due to Imperfection in Compression Paddle 2.8.1 Background ●

The radiologists noticed a small speck-like artifact within the mammogram, with microcalcification-like appearance, that was seen in most images from a particular unit.

●

A small speck (< 1 mm in diameter) can be seen with use of the new compression paddle.

2.8.3 Discussion The artifact is only seen in the patient images for which this particular compression paddle was used. The artifact does not appear in the same location because its projection shifts across the detector as the compression paddle height is adjusted according to breast thickness (▶ Fig. 2.8b).

A compression paddle had a crack and was replaced with a new paddle a few days prior.

2.8.2 Findings ●

●

A small artifact is seen (▶ Fig. 2.8a). It is visible in most but not all patient images, and at different locations.

2.8.4 Resolution An imperfection on the compression paddle manifests as a high-contrast microcalcification-like appearance in patient images. If cleaning the paddle does not remedy the artifact, the compression paddle should be replaced.

Fig. 2.8 (a) Location of the artifact in two mammograms with different compressed breast thicknesses. The red line indicates the location of the same detector row in both images. (b) The projected location of the artifact changes with compression paddle height.

37

| 09.11.19 - 01:52

Mammography

2.9 Case 9: Patient Motion Causing Blurred Parenchymal Structure in a Mammogram

●

Original image: compressed breast thickness was 72 cm, with 14 lb compression force. Repeat image: compressed breast thickness was 60 cm, with 23 lb compression force.

2.9.1 Background ●

This left-medial lateral oblique (LMLO) mammogram is part of a screening exam

●

of an asymptomatic patient. Breast cancer screening exams are rated using the BI-RADS scale. This exam resulted in a

●

BI-RADS score of 0 (recall) for technical reasons. The patient returned to the clinic and a technical repeat of this view was performed.

2.9.2 Findings ●

The parenchymal structure of the breast appears

●

blurred due to patient motion. Microcalcification clusters, which represent an important early indicator of breast cancer, might be missed due to insufficient spatial resolution

●

of this image. Exposure times in mammography are relatively long. This image was acquired with a 1.5-second X-ray exposure.

2.9.3 Discussion Mammography requires high spatial resolution in order to depict clusters of microcalcifications. Individual microcalcifications of clinical relevance have diameters ranging from 100 to 500 μm. In mammography, X-ray images are acquired with the breast under compression with typical compression forces between 25 and 45 lb. The technologist has to balance patient comfort and image quality. Too much compression force leads to patient discomfort. Too little compression force reduces breast immobilization and can result in patient motion, as seen in this case.

2.9.4 Resolution A repeat mammogram was obtained (▶ Fig. 2.9b). The technologist applied a higher compression force, resulting in better breast immobilization and the absence of motion artifacts.

Fig. 2.9 (a) Left-medial lateral oblique mammogram with signs of patient motion. The image appears unsharp. Fine detail of the parenchymal structure is blurred due to motion. (b) Repeated image with increased compression. Fine linear details in the parenchymal structure are visible in this image.

38

| 09.11.19 - 01:53

2.10 Case 10: EMI Artifact due to LVAD Device

2.10 Case 10: EMI Artifact due to LVAD Device 2.10.1 Background A patient received a routine mammogram. Due other health reasons, the patient has a left-ventricular assist device (LVAD).

2.10.2 Findings The mammogram (▶ Fig. 2.10) exhibits highfrequency striping near the chest wall edge. Visually, this artifact has similarity with grid artifacts that can be observed in radiography.

2.10.3 Discussion This EMI artifact occurs when electromagnetic fields emitted by the LVAD motor interfere with the detector readout electronics.

2.10.4 Resolution The radiologist needs to be aware that the presence of the LVAD devices in patients can cause such artifacts. Unless the mammography equipment manufacturer has implemented solutions into their imaging system to prevent such interference through shielding or image processing designed to remove these structures, these artifacts need to be tolerated.

Fig. 2.10 (a) Mammogram with the left-ventricular assist device (LVAD) device partially visible. When zoomed into a region near the chest wall (b), horizontal lines can be seen that are caused by the electromagnetic interference between the LVAD motor and the detector readout electronics.

39

| 09.11.19 - 01:53

Mammography

2.11 Review Questions 2.11.1 Case 1: Magnification Imaging 1. A small microcalcification (diameter = 100 μm) is imaged in contact mode. What is the approximate diameter of the microcalcification in the image? Assume dmag = 67.5 cm and ddet = 70 cm. a) 100 µm b) 200 µm c) 400 µm d) 500 µm 2. A small microcalcification (diameter = 100 μm) is

2.11.3 Case 3: X-ray Acquisition Technique Factors in Mammography 5. What is the most likely range of X-ray tube potentials used in mammography? a) 15–25 kV b) 25–35 kV c) 35–45 kV d) 45–50 kV 6. Subject contrast in a mammogram is most affected by which of these parameters? a) Milliampere-time product (mAs)

imaged in magnification mode. The detector

b) X-ray tube kilovoltage (kV)

pixel size is 90 μm. What magnification is

c) Exposure time

required so that diameter of the microcalcifica-

d) Focal spot size

tion in the image is twice the pixel size of the detector? a) 1 b) 1.8 c) 2 d) 2.3

2.11.4 Case 4: Digital Breast Tomosynthesis: Artifacts due to High-Contrast Objects 7. The resolution in digital breast tomosynthesis images is isotropic, similar to computed

2.11.2 Case 2: Focal Spot Size Selection in Magnification Views

tomography images. a) True b) False

3. In magnification imaging, a small focal spot is used to eliminate which of the following?

8. The origin of the artifacts from high-contrast

a) Scattered radiation

objects in digital breast tomosynthesis is most

b) Patient motion

similar to:

c) Focal spot blur

a) Detector pixel dropout artifact in conven-

d) Geometric distortion

tional mammography and radiography b) Artifacts from metal implants in

4. How wide is the blur from a 0.3-mm focal spot when the magnification factor is 2? a) 0.15 mm b) 0.3 mm c) 0.6 mm d) 0.8 mm

40

radiography c) Streaking artifacts from metal implants in computed tomography

| 09.11.19 - 01:53

2.11 Review Questions

2.11.5 Case 5: Effect of Image Post-Processing on the Appearance of a Mammogram 9. The effect of a high-pass filter is to:

14. If a small speck-like artifact is observed in the artifact image during weekly quality control testing, which actions should be taken? a) Clean all surfaces b) Rotate phantom

a) Increase image resolution

c) Gain calibration

b) Emphasize large detail in the image

d) Dead pixel mapping

c) Reduce noise

e) Replace detector

d) Emphasize small detail in the image

a) Spatial resolution

2.11.7 Case 7: Microcalcificationlike Appearance Caused by a Detector Artifact

b) Displayed contrast resolution

15. Which of the following is correct concerning

10. Unsharp masking is a post-processing algorithm that can improve:

c) Image noise

artifacts from dead pixels in the detector?

d) Geometric distortion

a) Dead pixels always appear as black spots. b) Dead pixels artifacts are always located at

2.11.6 Case 6: Artifact due to Detector Row Dropout 11. What is a likely cause for a partial detector row dropout? a) Dead pixels are lined up perfectly along a detector row. b) The image receptor consists of multiple tiled flat-panel detector elements. c) The detector electronic readout process is being interrupted. d) The detector panel has been physically damaged. 12. In tomosynthesis, a detector row dropout

identical image coordinates. c) The detector is free of dead pixels if a uniform test image does not show any artifacts. d) An artifact in a uniform test image always requires the detector to be replaced. 16. Which remedial actions should be considered when a detector artifact has been identified? a) Detector replacement b) Gain calibration c) Dead pixel mapping d) All of the above

FFDM mammogram.

2.11.8 Case 8: Artifact due to Imperfection in Compression Paddle

a) True

17. What is the most likely cause of a high-contrast

artifact looks the same in the synthetic 2D mammogram as it would in a conventional

b) False

small speck-like image artifact at a fixed location, which persists while imaging patients?

13. Prior to replacing the detector, which steps can

a) Bad pixels in detector

be taken to resolve a small speck-like image arti-

b) Dust during gain calibration

fact?

c) Imperfection in the compression paddle

a) None

d) Dust on breast support

b) Clean all surfaces c) Gain calibration

18. Compression force is applied

d) Dead pixel mapping

a) To reduce breast thickness

e) All of the above except choice a

b) To reduce the potential for patient motion

41

| 09.11.19 - 01:53

Mammography c) To reduce average glandular dose

c) Connection to external battery pack

d) All of the above

d) Abandoned cardiac leads

2.11.9 Case 9: Patient Motion Causing Blurred Parenchymal Structure in a Mammogram

Equations

19. What is the maximum allowable compression

a distance ddet between the focal spot and the

force in mammography?

Object magnification factor M, for an object positioned at a distance of dmag from the focal spot, and detector:

a) 25 pounds

M ¼ d det =d mag

b) 35 pounds c) 45 pounds

Diameter of the image of an object (xmag), imaged

d) 55 pounds

with magnification factor M, when the actual size of the object is x:

20. Which of the factors below is generally not a

x mag ¼ M  x

cause for patient motion? a) Breast compression force too low

Width of focal spot blur (wblur) for a focal spot of

b) Long exposure time

size w:

c) Magnification imaging d) Use of an X-ray grid

2.11.10 Case 10: EMI Artifact due to LVAD Device 21. Which of the following statements is correct? a) EMI artifacts can occur in computed radiography. b) EMI artifacts only occur in digital mammography detectors because of the high resolution. c) The appearance of an EMI artifact is similar to a grid artifact. d) An EMI artifact is a permanent detector artifact. 22. EMI artifacts occur due to the a) Presence of metal in the LVAD b) Spinning motor in the LVAD

42

w blur ¼ w  ðddet  dmag Þ=dmag ¼ w  M  1Þ

References [1] Villafana T. Generators, X-ray tubes, and exposure geometry in mammography. Radiographics. 1990; 10(3):539–554 [2] Pisano ED, Cole EB, Hemminger BM, et al. Image processing algorithms for digital mammography: a pictorial essay. Radiographics. 2000; 20(5):1479–1491 [3] Seeram E, Seeram D. Image postprocessing in digital radiology—a primer for technologists. J Med Imaging Radiat Sci. 2008; 39(1):23–41 [4] Yaffe MJ, Rowlands JA. X-ray detectors for digital radiography. Phys Med Biol. 1997; 42(1):1–39 [5] Ayyala RS, Chorlton M, Behrman RH, Kornguth PJ, Slanetz PJ. Digital mammographic artifacts on full-field systems: what are they and how do I fix them? Radiographics. 2008; 28(7): 1999–2008

| 09.11.19 - 01:53

3 Computed Tomography Karen L. Brown and Jason R. Gold

Introduction In computed tomography (CT) imaging, the X-ray tube and detector array rotate around the patient in a fixed geometry generating thousands of attenuation measurements through the patient volume of interest. The attenuation measurements are reconstructed into axial images which can then be reformatted into sagittal and coronal planes. The grayscale value displayed in each pixel represents the CT number in Hounsfield units of the corresponding voxel volume. CT number is a normalized measurement of the linear attenuation coefficient measured in the voxel relative to that of water,

defined to be zero in value. As such, tissues that are more attenuating than water will have positive values in a CT image, and tissues that are less attenuating than water will have negative values. Each type of tissue in a CT image has a specific Hounsfield unit range which may be used quantitatively to characterize tissues or pathology. Image quality in CT is affected by many parameters, some of which are selected by the operator prior to the acquisition, while others are selected prior to reconstruction of the data into tomographic planes. The same acquisition data can, therefore, be reconstructed in multiple ways to achieve different image quality end points dependent upon the clinical goals of the study.

43

| 09.11.19 - 01:53

Computed Tomography

3.1 Case 1: Ring Artifact 3.1.1 Background ●

Patient with a history of nephrolithiasis presented for CT of abdomen and pelvis examination.

●

A helical acquisition was acquired using a detector configuration of 24 × 1.2 mm channels.

●

Axial images of 3.0 mm thickness were reconstructed using a soft-tissue filter.

3.1.2 Findings A partial ring artifact is centrally located in all axial reconstructions.

3.1.3 Discussion The X-ray tube and detector array rotate around the CT gantry in a fixed geometry. The detector array consists of multiple detector channels in the z-direction, each containing many hundreds of individual detector elements in the x/y direction (▶ Fig. 3.1b).1 Each detector within the array measures the residual X-ray signal through the patient. If an individual detector within the array is not properly calibrated with respect to the

other detectors, or fails completely, artifacts will appear in the reconstructed images. In helical acquisitions, the artifact will appear as a partial ring (▶ Fig. 3.1a) and appear to rotate around isocenter throughout the imaging volume. Full ring artifacts will be present in axial scan acquisitions (▶ Fig. 3.1c). The artifact will be limited to the images corresponding to the affected detector channel. For example, if a 16channel system has a defective detector in channel 1, a full ring artifact will appear in the first reconstructed image and then again in image 17 but will not appear in images 2 to 16, assuming the width of the reconstructed slice is equal to the width of the detector channel.

3.1.4 Resolution Axial acquisitions of a uniform water phantom, reconstructed at the thinnest possible slice thickness, should be acquired and evaluated for ring artifacts by the CT technologist on a daily basis.2 When ring artifacts are identified, some CT systems will provide the user a method to recalibrate the detectors (often called an air calibration scan). If the air calibration scan is not available, or does not resolve the ring artifact, service personnel should be contacted and corrective maintenance be performed.

Fig. 3.1 (a) Partial ring artifact on helically acquired abdomen–pelvis computed tomography examination. (b) Multichannel CT detector array. (c) Full ring artifact on axially acquired phantom image.

44

| 09.11.19 - 01:53

3.2 Case 2: Effect of Patient Size on CT Number Accuracy

3.2 Case 2: Effect of Patient Size on CT Number Accuracy 3.2.1 Background ●

Bariatric patient with a history of renal calculi presents with abdominal pain.

●

A helical abdomen/pelvis CT is performed.

●

Incidental finding in the adrenal gland with elevated CT number measurement.

typically sized patient, this beam hardening effect is corrected for by the scanner. In bariatric patients, the beam hardening effect is more significant and can affect the accuracy of displayed CT numbers. When the patient’s tissues fall outside of the scan field of view (FOV), the system overestimates the attenuation provided by the tissues within the FOV. This is often referred to as a truncation artifact and appears as bright areas in the image (▶ Fig. 3.2a). Truncation artifact can also affect the accuracy of CT numbers used for quantitative analysis.3

3.2.2 Findings Use of CT number for quantitative analysis can be compromised in bariatric patients due to beam hardening and truncation artifact.

3.2.3 Discussion The X-ray tube used in CT produces a polyenergetic X-ray beam. Filters are placed at the exit port of the beam to remove low-energy X-rays increasing the average energy of the X-ray beam. A process called beam hardening. As the X-ray beam enters the patient, beam hardening continues as the lower-energy X-rays in the beam are attenuated with higher probability. The displayed CT number is a relative measure of attenuation as compared to the attenuation of water. As the beam becomes more energetic (hardened), less attenuation occurs in a given tissue which changes the CT number. In a

3.2.4 Resolution Some systems may offer an extended FOV reconstruction option as shown in ▶ Fig. 3.2b. In this case, the visual appearance of the truncation artifact was reduced by reconstructing using an extended FOV; however, little effect on measured CT number in the adrenal gland was realized. Patient positioning can also have a significant impact on beam hardening and truncation artifacts in bariatric patients. The patient in this case, presented for another scan 2 weeks later (▶ Fig. 3.2). Note the difference in the patient’s apparent shape and diameter. This was due to more effective wrapping of extraneous tissues by the technologist prior to scanning and is not related to patient weight loss. No truncation artifact is present and the change in measured Hounsfield units is significant.

Fig. 3.2 (a) Abdomen-pelvis computed tomography showing truncation artifact (arrows) and artificial elevation of measured CT numbers. Acquisition parameters: 120 kV, 557 mAs. (b) Extended field-of-view reconstruction of image shown in (a). (c) Follow-up abdomen-pelvis CT of patient. Acquisition parameters: 120 kV, 414 mAs. All other technique and reconstruction parameters were consistent with image shown in (a).

45

| 09.11.19 - 01:53

Computed Tomography

3.3.2 Findings

contrast of the seminal vesicle structure shown in ▶ Fig. 3.3a which was acquired at 100 kV as compared to the same structure in ▶ Fig. 3.3b acquired at 120 kV. Tube voltage (kV) also affects the efficiency of X-ray production in the X-ray tube. At a lower kV setting, the number of X-rays produced is reduced by the ratio of the change in kV squared to cubed (ΔkV2–3). Quantum noise increases when fewer X-rays are used to make the image, which will have a negative effect on the visibility of low-contrast structures. To compensate, a decrease in kV is often accompanied by an increase in tube current (mA). As shown in ▶ Fig. 3.3a, b, an appropriate increase in mA to achieve an acceptable level of quantum noise levels can be achieved at a significant dose reduction.

Contrast is improved at the lower kV setting at a dose savings of approximately 17%.

3.3.4 Resolution

3.3 Case 3: Effect of kV Selection on Image Quality and Dose 3.3.1 Background ●

Patient with a history of cecal and appendiceal adenocarcinoma with multiple prior CT examinations presents for chest/abdomen/pelvis CT with contrast post resection with peroneal nodules for evaluation of treatment.

●

A helical acquisition is acquired at reduced kV compared to prior studies.

3.3.3 Discussion Lowering tube voltage (kV) produces an X-ray beam with lower average and maximum energy. Lower X-ray energies will be attenuated in a given tissue more than higher-energy X-rays. This is primarily governed by differences in attenuation that occurs due to photoelectric absorption which is approximately proportional to the atomic number (Z) cubed of the tissue, and inversely proportional to the energy (E) cubed of the X-ray beam. As such, the relative attenuation between two tissues increases at lower kV settings. On a CT image, this results in a greater difference in Hounsfield units between the two tissues, and therefore, greater contrast. Note the increased

Tube voltage and current settings should be adjusted to optimize image quality and dose. The usefulness of this technique will depend on patient size and the anatomy/pathology of interest. A lower kV setting can often be used on smaller patients. In very large patients, the mA limitations of the X-ray tube may result in increased photon starvation artifacts. Beam hardening artifacts will also be enhanced at low kV settings. Lower-energy beams are also effective when using iodinated contrast agents as the average beam energy more closely aligns with the k-edge absorption peak of iodine. Care should be taken when using CT numbers for quantitative analysis as the CT number is affected by the kV setting.

Fig. 3.3 (a) Image acquired at 100 kV and 200 mAs with a displayed volume CT dose index (CTDIVOL) of 8.37 mGy. (b) Image acquired at 120 kV and 140 mAs with a displayed CTDIVOL of 10.06 mGy.

46

| 09.11.19 - 01:53

3.4 Case 4: Image Quality Variation with Reconstructed Slice Thickness

3.4 Case 4: Image Quality Variation with Reconstructed Slice Thickness 3.4.1 Background ●

Patient presents to the emergency department following a motor vehicle collision.

●

CT thorax, CT abdomen, and pelvis examinations following the administration of contrast were acquired.

●

Axial images of 3.0 and 1.5 mm slice thickness were reconstructed.

3.4.2 Findings Small fracture of the clavicle is not visible with 3.0 mm reconstructed slices.

3.4.3 Discussion Reconstructed slice thickness affects the level of quantum noise, spatial resolution, and partial volume averaging present in CT images. Slice thickness affects the size of the reconstructed tissue voxel. Each voxel of tissue is displayed as one shade of gray in the corresponding image pixel. When the voxel contains more than one type of tissue or pathology, the CT number displayed

represents the average attenuation measurement of all the tissues within the voxel. This principle is called partial volume averaging and is inherent to the CT image reconstruction process. As the tissue voxel becomes larger and there is more signal averaging, there is less spatial differentiation of the signal in a given direction and the image appears more blurred, with lower spatial resolution. ▶ Fig. 3.4a is reconstructed with a 1.5 mm slice thickness and ▶ Fig. 3.4b is reconstructed with a 3.0 mm slice thickness. The thinner reconstructions are sharper (better spatial resolution) and have less partial volume averaging. Note the subtle clavicle fracture seen on the 1.5-mm reconstructions which is not visible on the 3.0-mm reconstructions. Quantum noise is also affected by the size of the reconstructed voxel. The amount of signal (number of X-rays) interacting within a given voxel is directly proportional to voxel size. Quantum noise changes inversely with the square root of the change in signal.

3.4.4 Resolution Voxel size is a function of the pixel size and the reconstructed slice thickness selected by the operator. When reconstructed slice thickness is increased, more partial volume averaging occurs in the slice thickness direction. When a thinner reconstructed slice thickness is selected, there is better differentiation of tissues along the slice

Fig. 3.4 (a) Computed tomography (CT) image reconstructed with 1.5 mm slice thickness. (b) CT image reconstructed with 3.0 mm slice thickness.

47

| 09.11.19 - 01:53

Computed Tomography thickness direction resulting in better spatial resolution and less partial volume averaging. CT protocols often include multiple reconstructions of the same acquisition at different reconstructed slice thicknesses to provide the clinician with high spatial resolution image series as well as low noise image series for evaluation.

48

The only other parameter that affects the size of the reconstructed voxel in CT is the FOV selected by the operator prior to reconstruction. FOV is often selected to encompass the anatomy of interest and determines the size of the tissue voxel and corresponding image pixel in the x and y dimension.

| 09.11.19 - 01:53

3.5 Case 5: Image Quality Variation with Reconstruction Filter

3.5 Case 5: Image Quality Variation with Reconstruction Filter 3.5.1 Background ●

Patient with a history of bacterial meningitis and multiple brain abscesses presents for follow-up CT to evaluate the response to treatment with IV antibiotics.

●

Helical CT of the head is acquired with contrast and reconstructed using filtered back projection.

●

A reduced mA technique is implemented in an

signal intensity. The tube current (mA) used during the acquisition of the exam affects the total number of X-rays produced. If the mA is reduced by one half, signal intensity is reduced by one half. They are directly proportional. The level of quantum noise in the image will increase as 1/√ (change in signal) or 1/√1/2 (approximately 40%). In this case, the images reconstructed using filtered back projection (▶ Fig. 3.5a) are very noisy. The radiologist is concerned that important findings may be missed as increased levels of quantum noise diminish visibility of low-contrast structures. The advantage of reduced mA techniques is lower patient dose. Patient dose is directly proportional to the mA setting selected by the technologist.

effort to reduce patient radiation dose.

3.5.2 Findings The reconstructed images have elevated quantum noise due to the use of a reduced mA technique.

3.5.3 Discussion The level of quantum noise in a CT image is related to 1/√N, where N is equal to the number of X-ray photons used to generate the image, also called the

3.5.4 Resolution There are several acquisition and reconstruction parameters that affect the level of quantum noise in an image. One of the reconstruction parameters selectable by the operator is the reconstruction filter or kernel. The reconstruction filter will affect how much smoothing out of the noise occurs in the image. A soft tissue or standard filter will have lower image noise compared to a sharp filter, such as a bone or lung filter, at the expense of lower

Fig. 3.5 (a) Filtered back projection reconstruction using a smooth filter with a window width/window level setting of 100/40. (b) Filtered back projection reconstruction using a sharp filter with a window width/ window level setting of 100/40. (c) Filtered back projection reconstruction using a sharp filter with a window width/window level setting of 2500/500. (d) Iterative reconstruction using a smooth filter with a window width/window level setting of 100/40.

49

| 09.11.19 - 01:53

Computed Tomography spatial resolution. The image in ▶ Fig. 3.5a was reconstructed using a soft-tissue (smooth) reconstruction filter. At the same window/level setting, the same data reconstructed with a sharp filter (▶ Fig. 3.5b) has much higher noise but spatial resolution has improved. Sharp reconstructions are often viewed on a different window/level setting to emphasize the spatial information in the structures of interest (▶ Fig. 3.5c). The type of reconstruction algorithm selected by the operator will also have an effect on image noise.

50

Iterative reconstruction techniques are available on all modern CT scanners and produce images with lower noise than filtered back projection techniques.4 ▶ Fig. 3.5d shows the same image reconstructed using an iterative technique. Note the significant reduction in quantum noise compared to ▶ Fig. 3.5a. Spatial resolution in the iterative reconstruction is also maintained which is another advantage of this technique. Iterative reconstruction is often used with reduced mA or kV techniques to overcome increases in quantum noise.

| 09.11.19 - 01:53

3.6 Case 6: Displayed Volume CT Dose Index and Patient Size

3.6 Case 6: Displayed Volume CT Dose Index and Patient Size 3.6.1 Background ●

A 6-year-old patient with medulloblastoma presents for CT of the thorax with contrast for evaluation of possible pulmonary embolus.

●

Patient has received multiple CT examinations over the past year.

3.6.2 Findings Volume CT dose index (CTDIVOL) and dose length product (DLP) displayed on the dose summary page underestimate dose to the patient.

3.6.3 Discussion Patient dose is dependent on the radiation output of the CT scanner and patient size. The CTDIVOL (in units of mGy) displayed on the CT scanner is a measure of the radiation output as estimated to one of two-sized polymethyl methacrylate phantoms, a small 16-cm-diameter phantom or a large 32-cm-diameter phantom.5 When patient size varies from the size of the phantom used, the CTDIVOL displayed on the CT scanner may over- or underestimate patient dose. In the case presented, the dose summary report (▶ Fig. 3.6a) estimates 2.08 mGy as the CTDIVOL to the large 32-cm phantom from the chest scan. The actual diameter of the patient, as shown in ▶ Fig. 3.6b is significantly less than 32 cm. In this case, the CTDIVOL underestimates the dose to the patient which would be higher than the value displayed on the dose summary page. When patient size is larger than the phantom indicated on the dose summary page, patient dose is overestimated and would be lower than the displayed CTDIVOL value. The DLP is calculated by multiplying the estimated CTDIVOL by the scan length and has units of

mGy-cm. DLP provides an estimate of the total energy imparted to the scan volume.5 Any error in estimating patient dose using CTDIVOL is propagated in the calculation of DLP. The use of dose values displayed on the CT scanner to estimate patient dose can be particularly problematic for chest, abdomen, and pelvic examinations of pediatric patients. The displayed dose for body scans is typically estimated using the large 32-cm phantom (although some older systems may use the 16-cm phantom) which may underestimate dose in pediatric patients by a factor of 2 or 3. For this reason, care must be taken when using displayed dose metrics for benchmarking and protocol optimization.

3.6.4 Resolution When it is necessary to estimate patient dose from a CT examination, the clinician should be aware of the limitations of displayed or reported CT dose metrics on the scanner. The American Association of Physicists in Medicine has developed a method to provide a better estimate of patient dose.6 The size-specific dose estimate (SSDE) is calculated by multiplying the displayed CTDIVOL by a correction factor that accounts for the diameter of the patient as compared to the diameter of the phantom used to estimate CTDIVOL. SSDE is not currently displayed on CT scanners or on the dose summary page. Some third-party dose management software programs do provide SSDE calculations that can be compared to national benchmarks such as the American College of Radiology Dose Index Registry. If these tools are not readily available to the clinician, a medical physicist should be consulted to provide a patient dose estimate. As a general rule of thumb, if the large (32 cm) phantom is used by the scanner to estimate CTDIVOL, the dose to a 16-cm-diameter patient will be approximately two times the value displayed on the scanner. The estimated SSDE for the patient presented in this case is 3.6 mGy.

Fig. 3.6 (a) Computed tomography dose summary page. (b) Axial CT scan used to estimate patient diameter.

51

| 09.11.19 - 01:53

Computed Tomography

3.7 Case 7: Beam Hardening Artifact 3.7.1 Background Patient with suspected subdural hematoma presents for head and neck CTA with and without contrast.

3.7.2 Findings Beam hardening artifacts are observed in contrastenhanced images.

3.7.3 Discussion Beam hardening occurs when the polyenergetic X-ray beam passes through material.7 Higher density and higher atomic number materials such as metal, bone, and contrast agents preferentially attenuate lower-energy X-rays resulting in a higher average energy X-ray beam exiting the material. Because, the higher-energy beam has greater penetrating ability, tissues that lie between, or are adjacent to these structures appear to be less attenuating. In the reconstruction, lower CT numbers than what actually represent the tissue are calculated and displayed. This results in shadowing or dark bands as shown in ▶ Fig. 3.7a. The magnitude of beam hardening that occurs will depend on the path length of the

X-ray beam through these dense objects. Note the presence of beam hardening around the bone structures in the noncontrast image (▶ Fig. 3.7b). In the contrast image, beam hardening artifact is more noticeable when the beam passes through all three contrast-filled structures but is not apparent at other angular pathways where the beam passes through a shorter axis of each individual contrast-filled structure.

3.7.4 Resolution Beam hardening artifacts can be minimized with the use of higher-energy X-ray beams. Increasing the tube voltage (kV) produces a higher-energy X-ray beam but this will also affect image contrast and patient dose. Dual energy CT acquisitions acquire data at two separate beam energies. A low kV setting, such as 80 kV, and a high kV setting, such as 140 kV, are typical. The attenuation difference between tissues will be different at the low kV setting compared to the high kV setting. This information can be used to reconstruct virtual monoenergetic images at higher energy, eliminating the effect of low-energy X-rays on beam hardening.8,9 Angulation of the gantry to avoid passage of the beam through the long axis of known, high-density structures will help mitigate the artifact. This option is not available on all CT scanners. Iterative reconstruction techniques can also be used to minimize the appearance of beam hardening artifacts.

Fig. 3.7 (a) Contrast scan showing shading artifact from beam hardening when the angular orientation of the beam passes through multiple contrast-filled structures. (b) Noncontrast scan shows less significant beam hardening artifact when the beam passes through the long axis of bone structures.

52

| 09.11.19 - 01:53

3.8 Case 8: Partial Volume Artifact

3.8 Case 8: Partial Volume Artifact 3.8.1 Background ●

ICU patient presents with history of right thalamic hemorrhage and right frontal contusion.

●

Portable axial CT scan of the head without contrast with a reconstructed slice thickness of 5 mm is performed.

3.8.2 Findings Hyperdensity in right frontal lobe due to partial volume averaging is not indicative of contusion per history.

3.8.3 Discussion Each pixel in a CT image displays a single shade of gray representing the calculated CT number for the associated voxel of tissue. When more than one type of tissue is contained within a given voxel, the average attenuation value of the tissues within the voxel is used to calculate the CT number and corresponding displayed grayscale value. Head CT scans are commonly reconstructed with a 5 mm slice thickness, resulting in relatively large voxels in the z-direction (through the scanner bore). ▶ Fig. 3.8a shows a hyperdensity in the right frontal lobe region indicated by

the arrow. The increased signal is due to the averaging of skull and brain matter attenuation within the voxel (▶ Fig. 3.8b) resulting in elevated CT number and brighter shade of gray presentation in the image. Follow-up CT (▶ Fig. 3.8c) shows no hyperdensity due to contusion and that the elevated CT numbers are likely due to partial volume averaging.

3.8.4 Resolution The magnitude of partial volume artifact is primarily affected by reconstructed slice thickness. Slice thickness controls the z-dimension of a tissue voxel. It is common for most CT examinations to be reconstructed using thick and thin reconstructions. Partial volume averaging will be diminished in thinner slice images compared to thick image reconstructions. The x–y dimension of the voxel is equal to the size of the image pixel which is determined by the FOV divided by the size of the image matrix. For most modern CT scanners, the matrix size is fixed at 512 × 512 pixels. Adjusting the FOV is, therefore, the single parameter adjustment that can affect pixel size. The size of each pixel in CT is typically much smaller than the slice thickness, so changing FOV has less effect on partial volume averaging than slice thickness. Reconstructing and viewing images in different reconstruction planes may also affect partial volume averaging as the tissue types within a given voxel may vary along the direction of reconstruction.

Fig. 3.8 (a) Axial computed tomography of the brain reconstructed with 5 mm slices shows hyperdensity in the right frontal lobe. (b) Scout image showing location of adjacent axial slices (white lines) and voxel encompassing both skull and brain matter. Note: for illustrative purposes only, voxel size and position are not accurately depicted. (c) Axial followup CT shows no hyperdensity in the right frontal lobe indicating the effect on prior image was likely due to partial volume averaging.

53

| 09.11.19 - 01:53

Computed Tomography

3.9 Case 9: Metal Artifact

3.9.4 Resolution

3.9.1 Background

There are several techniques to minimize the appearance of metal artifacts.11 Thin acquisition sections are recommended to reduce partial volume averaging; however, larger reconstructed slice thickness averages the noise over larger voxels and mitigates the appearance of streaking. ▶ Fig. 3.9a was reconstructed with 3.0 mm slice thickness and shows mild reduction in streak artifact compared to ▶ Fig. 3.9b which was reconstructed with 1.0 mm slice thickness. Higher tube voltage (kV) increases the penetrating ability of the X-ray beam through metal structures and also increases the number of X-ray photons in the beam helping to overcome both beam hardening and photon starvation effects. Note significant reduction in metal streak artifact seen in ▶ Fig. 3.9c acquired at 140 kV. Dual energy CT can be used to generate virtual monoenergetic images.89 Using this technique, low-energy X-rays in the beam that contribute significantly to beam hardening are removed, minimizing their contribution to streak artifacts. Metal artifact reduction software is available from equipment and third party vendors. The use of iterative reconstruction techniques reduces noise, minimizing artifact intensity. These techniques can be used separately or in combination to reduce the appearance of metal streak artifacts.

●

Patient with possible leg cellulitis and fluid collection in the anterolateral leg is referred for CT.

●

Helical CT scan of the lower leg with 1.0- and 3.0-mm reconstructions is acquired.

3.9.2 Findings Metal artifact precludes visualization of possible soft tissue fluid collection.

3.9.3 Discussion The presence of metal within the CT scan FOV causes severe streaking artifact. There are several phenomenon contributing to the appearance of streaking including beam hardening and photon starvation effects. In addition, the measured linear attenuation values of metal structures may be beyond the dynamic range of the CT system. Visualization of anatomy and pathology in the vicinity of metal material may be significantly compromised as a result. In this case, the radiologist was unable to make a definitive diagnosis due to the severe streaking artifact (▶ Fig. 3.9a, b) in proximity to the structures of interest.

Fig. 3.9 (a) Metal artifact reconstructed with 3.0 mm slice thickness showed mild improvement of streak artifact compared to 1.0 mm reconstruction as shown in (b). (b) Image reconstructed with 1.0 mm slice thickness shows slightly enhanced streaking from metal artifact compared to 3.0 mm reconstruction as shown in (a). (c) Image acquired at 140 kV shows significant reduction in metal streak artifact.

54

| 09.11.19 - 01:53

3.10 Case 10: Motion Artifact

3.10 Case 10: Motion Artifact 3.10.1 Background ●

Patient presents with left-sided weakness.

●

Noncontrast helical CT is performed with sagittal and coronal reformats.

3.10.2 Findings Evaluation of the study was compromised due to motion artifact, requiring the technologist to repeat the scan.

3.10.3 Discussion The appearance of motion artifacts can be quite varied depending on the source of motion (patient, respiratory, cardiac, etc.) and severity. Respiratory motion may appear as a generalized loss of resolution in the anterior chest cavity with little effect in other areas of the image. As shown in ▶ Fig. 3.10, significant patient motion can cause streaking, shading, ghosting, and incongruence of anatomical features. In some cases, the degradation in image quality caused by motion requires the scan to be repeated resulting in increased radiation dose to the patient.

3.10.4 Resolution Patient motion can be mitigated with adequate exam preparations to include careful patient positioning and appropriate use of immobilization devices, explanation of the examination process to the patient prior to the scan, and provision of clear instructions throughout the scanning process. Optimization of scan parameters is also essential, when imaging anatomical features with higher probability of involuntary motion, gantry rotation time and helical pitch are adjusted accordingly to optimize scan acquisition time. Shorter gantry rotation times and higher helical pitch result in shorter scan acquisition times. For example, head scans typically

Fig. 3.10 Image presentation of significant patient motion during computed tomography acquisition.

incorporate longer gantry rotation times (0.8–2.0 seconds) as compared to abdominal scans (0.25–0.5 seconds) during which respiratory motion is more likely to occur. For patients with a high probability of motion, either due to their condition or age, sedation may be appropriate. Systems with larger beam widths and dual source technology also provide opportunities for faster scanning, minimizing the potential for motion artifacts. Cardiac motion creates a significant challenge when imaging heart structures and vasculature. Artifacts related to cardiac motion can manifest in a variety of forms including blurring, ghosting, and misregistration.12 Prospective or retrospective cardiac gating techniques are often employed to essentially “freeze” heart motion during the selected portion of the cardiac cycle. The effectiveness of these techniques is dependent on patient heart rate and stability. Beta blockers may be administered to the patient to decrease patient heart rate and extend the period of diastole when using cardiac gating techniques.

55

| 09.11.19 - 01:53

Computed Tomography

3.11 Review Questions

b) Bone

3.11.1 Case 1: Ring Artifact

d) Iodine

c) Fat

1. What reconstructed slice thickness will provide the greatest visibility of detector malfunctions

6. What CT acquisition parameter has the largest

in a CT scanner?

effect on quantum noise in the image?

a) 0.6 mm

a) mA

b) 1.2 mm

b) kV

c) 3.0 mm

c) Rotation time

d) 5.0 mm

d) Collimation

2. What is the purpose of the daily air calibration scan recommended by some CT scanner manufacturers? a) Assess the CT number accuracy of water and air b) Adjust the gain settings of individual detector elements c) Measure noise standard deviation of the system d) Disable malfunctioning detector channels

3.11.2 Case 2: Effect of Patient Size on CT Number Accuracy 3. What factor affects the measured CT number

3.11.4 Case 4: Image Quality Variation with Reconstructed Slice Thickness 7. By what factor does quantum noise in the image change when slice thickness is decreased from 3.0 to 1.5 mm? a) 0.75 b) 1.0 c) 1.4 d) 2.0 8. What is the primary advantage of thicker reconstructed slice thickness in CT? a) Improved spatial resolution

for a given tissue?

b) Faster acquisition times

a) Beam energy

c) Lower patient dose

b) Tube current modulation

d) Increased low-contrast visibility

c) Gantry rotation speed d) Window level/width 4. What affect does increasing the FOV have on image spatial resolution? a) Decreases

3.11.5 Case 5: Image Quality Variation with Reconstruction Filter 9. An increase in what parameter will reduce

b) Increases

quantum noise in a CT image with no direct

c) Stays the same

effect on patient dose? a) Pitch

3.11.3 Case 3: Effect of kV Selection on Image Quality and Dose 5. For a given X-ray beam energy, what material

56

b) Slice thickness c) Rotation time d) kV 10. What is the primary advantage of iterative

has the highest probability of attenuating an

reconstruction techniques as compared to

X-ray photon?

filtered back projection?

a) Air

a) Lower noise images

| 09.11.19 - 01:53

3.11 Review Questions b) Faster reconstruction time c) Reduced image storage requirements d) Minimal partial volume averaging

3.11.8 Case 8: Partial Volume Artifact 15. What effect will decreasing FOV have on

3.11.6 Case 6: Displayed Volume CT Dose Index and Patient Size

quantum noise when using fixed acquisition

11. How is the displayed CTDIVOL related to the tube

b) Decrease

current setting during a CT acquisition?

parameters? a) Increase c) No effect

a) Directly b) Inversely

16. What effect will decreasing reconstructed slice

c) Exponentially

thickness have on spatial resolution?

d) One over the square root

a) Increase b) Decrease

12. What is a typical CTDIVOL reported for a routine

c) No effect

adult abdomen/pelvis CT examination of average-sized patient? a) 1–5 mGy b) 10–15 mGy

3.11.9 Case 9: Metal Artifact 17. What is the dominant interaction of 80 kV

c) 25–30 mGy

X-rays in metal?

d) 45–50 mGy

a) Coherent scattering b) Compton scattering

3.11.7 Case 7: Beam Hardening Artifact 13. What component of CT imaging system

c) Pair production d) Photoelectric absorption 18. What quantity is used to calculate CT number

corrects for variations in X-ray beam intensity

(Hounsfield unit)?

due to different X-ray path lengths through

a) Linear attenuation coefficient

the patient?

b) Mass attenuation coefficient

a) Anode

c) Tissue atomic number

b) Bow-tie filter

d) Tissue density

c) Collimator d) Grid 14. An increase in what parameter will reduce the

3.11.10 Case 10: Motion Artifact 19. What is the effect of using a shorter gantry

difference in grayscale values between adjacent

rotation time, assuming all other protocol

tissues in a CT image.

parameters remain unchanged?

a) mA

a) Decreased spatial resolution

b) kV

b) Longer scan acquisition time

c) Rotation time

c) Increased quantum noise

d) Pitch

d) Higher patient dose

57

| 09.11.19 - 01:53

Computed Tomography 20. What is the effect of increasing helical pitch assuming all other protocol parameters remain unchanged? a) Improved visibility of low-contrast structures b) Lower patient dose c) Higher spatial resolution d) Decreased quantum noise

Equations Pixel size ¼

FOV Matrix Size

Voxel size ¼ pixel size  slice thickness CT # ðHUÞ ¼

 t  w  1000 w

μt is the average linear attenuation coefficient of the tissues within a voxel. μw is the linear attenuation coefficient of water

References [1] Bushberg JT, Seibert JA, Leidholdt E, Boone J. The Essential Physics of Medical Imaging. 3rd ed. Philadelphia: Wolters Kluwer Health/Lippincott Williams & Wilkins; 2012 [2] The American College of Radiology. Computed Tomography: Quality Control Manual, 2017

58

[3] Fursevich DM, LiMarzi GM, O’Dell MC, Hernandez MA, Sensakovic WF. Bariatric CT imaging: challenges and solutions. Radiographics. 2016; 36(4):1076–1086 [4] Padole A, Ali Khawaja RD, Kalra MK, Singh S. CT radiation dose and iterative reconstruction techniques. AJR Am J Roentgenol. 2015; 204(4):W384–92 [5] Report No AAPM. 96, The Measurement, Reporting, and Management of CT Dose, The American Association of Physicists in Medicine, 2008 [6] Report No AAPM. 204, Size Specific Dose Estimates (SSDE) in Pediatric and Adult Body CT Examinations, The American Association of Physicists in Medicine, 2011 [7] Barrett JF, Keat N. Artifacts in CT: recognition and avoidance. Radiographics. 2004; 24(6):1679–1691 [8] McCollough CH, Leng S, Yu L, Fletcher JG. Dual- and multi-energy CT: principles, technical approaches, and clinical applications. Radiology. 2015; 276(3):637–653 [9] Grajo JR, Patino M, Prochowski A, Sahani D. Dual energy CT in practice: basic principles and applications. Appl Radiol. 2016; 45(7):6–12 [10] Katsura M, Sato J, Akahane M, Kunimatsu A, Abe O. Current and novel techniques for metal artifact reduction at CT: practical guide for radiologists. Radiographics. 2018; 38(2): 450–461 [11] Lee MJ, Kim S, Lee SA, et al. Overcoming artifacts from metallic orthopedic implants at high-field-strength MR imaging and multi-detector CT. Radiographics. 2007; 27(3): 791–803 [12] Kalisz K, Buethe J, Saboo SS, Abbara S, Halliburton S, Rajiah P. Artifacts at cardiac CT: physics and solutions. Radiographics. 2016; 36(7):2064–2083

| 09.11.19 - 01:53

4 Magnetic Resonance Imaging Puneet Sharma

Introduction From its inception in the 1970s, and through humble beginnings in the early 1980s, magnetic resonance imaging (MRI) has experienced overwhelming growth in utility and innovation across virtually all diagnostic applications. The power of MRI lies in image contrast, exploiting the inherent magnetic properties of protons predominantly found in tissue water, but also in other biochemical species. The essential tool for an MRI experiment is a large main magnetic field (> 0.5 Tesla), which creates an observable net magnetization in biological samples that can be subsequently manipulated for image formation. A carefully designed collection of radiofrequencies (RF) and time-varying gradient pulses work in conjunction with the main magnetic field to produce image contrast “maps” from received signals that are “weighted” toward specific tissue properties, such as T1, T2, T2*, proton density (PD), and, if desired, diffusion and flow. This image contrast versatility is akin to histopathological tissue staining, wherein specific tissue substructures are highlighted by altering the chemical fixation. Anatomical visualization in MRI can be prescribed in any orientation using multiplanar 2D or 3D slices; however, gating and repetition can be employed to manifest a temporal dimension, allowing cine or 4D visualization. Since MRI acquisitions depend on tissue relaxation properties (i.e., T1 and T2), adequate time is needed to impart appropriate weighting to the encoded MR signal. In its basic sense, the received signal represents information pertaining to tissue properties, pulse timing, and signal location,

which is subsequently stored. A multitude of additional signals are generated, encoded, and stored that are eventually combined and decoded through a process called Fourier analysis to produce a final image. Though some MR acquisitions are on the order of 500 miliseconds for one slice, most scan times are on the order of minutes due to the repeated process of excitation, encoding, timing, and signal reception. Many significant innovations in MRI over the past two decades have focused on reducing acquisition times, mostly by limiting the required data needed for image reconstruction. In actuality, only a tiny percentage of protons make up the observable net magnetization for MRI. This places significant importance on coil reception sensitivity to optimize signal-to-noise ratio (SNR). Moreover, these SNR constraints may place practical limits on image resolution, especially in comparison to computed tomography (CT). Although SNR can be improved through signal averaging, this prolongs scan times significantly.

Common Image Quality Problems ●

Motion (bulk tissue movement and flow)

●

Susceptibility

●

Aliasing

●

Truncation

●

Field inhomogeneity

●

Coil use and placement

59

| 09.11.19 - 01:53

Magnetic Resonance Imaging

4.1 Case 1: Appearance of Discrete Image Ghosts on Abdominal Imaging 4.1.1 Background ●

Subject underwent routine abdominal MRI, using multislice gradient echo imaging.

●

This type of gradient echo data acquisition interleaves collection of phase-encoded data signals from all slices over the course of one repetition time (TR) period (~ 170 miliseconds).

4.1.3 Discussion The essence of MR signal acquisition and reconstruction relies on data consistency in k-space over the complete duration of the scan (seconds to minutes). A positional change of an object during acquisition will induce an amplitude or phase modulation of the expected k-space encoding step, especially in relation to previous (and future) encoding steps. This modulation in k-space will manifest as a replication of the object in the resultant image due to properties of the discrete Fourier transform. The nature of the replication depends on the nature of the object motion, and how it

Subsequent phase-encode steps are repeated every TR period, until k-space is filled. ●

Most abdominal MRI acquisitions, such as this, require the subject to suspend breathing for the duration of the scan. Typically, the breath-hold duration is < 20 seconds.

4.1.2 Findings ●

Subject was not able to suspend breathing during the acquisition.

●

Upon reconstruction, distinct image “copies” (ghosts) are seen propagating in the anteriorposterior direction (▶ Fig. 4.1, ▶ Fig. 4.2, and ▶ Fig. 4.3). Fig. 4.1 Axial T1 gradient echo of the abdomen showing gross appearance of image ghosts, propagating in the anterior-posterior direction.

Fig. 4.2 (a–c) Three abdominal cases, showing acquisition during (a) heavy breathing (period motion); (b) incomplete or irregular breath holding, and (c) perfect breath holding. Variation of phase encode data collection accompanies each case.

60

| 09.11.19 - 01:53

4.1 Case 1: Appearance of Discrete Image Ghosts on Abdominal Imaging Flow due to moving spins also may induce image ghosts. Like respiratory motion, flow pulsatility coupled with signal amplitude changes from unsaturated spins moving through the imaging plane and can cause modulation of k-space across phase-encode steps (▶ Fig. 4.3).

4.1.4 Resolution

Fig. 4.3 Discrete ghosts from pulsatile flow in the aorta propagate in the anterior-posterior direction.

synchronizes with data acquisition: periodic motion, such as respiration, will usually result in discrete ghosts, while random motion, such as eye movement and swallowing, will result in faint, unstructured ghosts. Since the effective sampling rate is slower in the phase encode direction (1/TR, for single-echo imaging; 1/echo-spacing, for echo train imaging) compared to the frequency encode direction (1/Δt = bandwidth (BW), Δt = sampling interval), the sensitivity of k-space modulation and inconsistency is commonly seen along the phase encode direction. This is depicted in ▶ Fig. 4.2 which plots the degree of (normalized) k-space modulation as a function of three types of motional behaviors: deep periodic breathing, shallow breathing, and perfect breath holding. As seen, the change in amplitude among phase-encode steps are much greater than individual frequency-encode steps.

The simplest tactic to remedy the clinical significance of image ghosts is to exchange phase and frequency directions in order to redirect ghosts into a perpendicular direction, and better reveal tissue under examination. However, any strategy to reduce or entirely eliminate image ghosts involves ensuring k-space consistency over the duration of the data acquisition. As seen in ▶ Fig. 4.2, eliminating ghosts ultimately requires synchronizing phase-encode data collection with known object motion. This synchronization, as employed with navigator or respiratory gating, collects data only from particular motional states (e.g., expiration) and disregards others. Depending on the complexity of motion, data synchronization may result in long scan times or even insignificant motion compensation. Alternatively, if one is able to reduce TR such that the phase encode sampling rate is high relative to the object motional rate, ghosts may become indistinguishable from edge blurring. Reducing TR and echo spacing may also allow breath holding. This latter strategy is the most robust compensation method and is becoming increasingly more applicable due to advancements in parallel imaging and compressed sensing.

61

| 09.11.19 - 01:53

Magnetic Resonance Imaging

4.2 Case 2: A Well-Defined Area of Signal Hyperintensity Appears Bilaterally at the Level of the Internal Auditory Canal on Diffusion-Weighted MRI, Affecting Visualization of Surrounding Structures 4.2.1 Background ●

Routine brain MRI exam without specific pathologic indication. Diffusion-weighted imaging (DWI) is a standard-of-care acquisition for describing cellular integrity of various tissue types.

●

DWI is an echo-planar imaging (EPI) technique, utilizing large directionally sensitive gradients to encode movements related to diffusion.

●

EPI is a rapid gradient echo method that collects all k-space data following one RF excitation

common outcomes: (1) image distortion (or warping), and (2) susceptibility related signal loss. When adjacent tissues have very different magnetic susceptibility, image distortion is likely since the local magnetic field is altered from its expected value. This inhomogeneous environment may disrupt the applied spatial encoding gradients during an MRI acquisition, which are assumed to be linear. In the frequency-encoding direction, a nonlinear gradient will change the overall frequency distribution of encoded spins, where some spatial locations may now be “mapped” with more than one spatial frequency (▶ Fig. 4.5), resulting in image distortion or signal “pile-up.” The more shallow the applied gradient (low encoding BW), the more significant the spatial mismapping and distortion. Even though this effect occurs predominantly in the frequency-encoding direction, singleshot EPI shows susceptibility-related distortion in the phase-encode direction (▶ Fig. 4.4) due to the relatively low sampling rate in this direction. Accelerated signal loss is another result of high magnetic susceptibility environments. The large local field

(“single-shot”). This typically involves fast gradient reversals, interleaved with incremental phase-encoding steps.

4.2.2 Findings ●

The high signal intensity is an artifact evolving from constructive signal “pile-up” related to susceptibility (arrows, ▶ Fig. 4.4).

●

The internal auditory canal (IAC) region is an air-filled region with changing geometry that represents a sharp transition of magnetic susceptibility compared to the rest of the brain structure.

4.2.3 Discussion Tissues and other substances in the body will alter the applied magnetic field based on their tissue properties and chemical composition. “Magnetic susceptibility” is a term that describes the degree to which a substance is able to disrupt the local magnetic field in terms of its strengthening (paramagnetic) or weakening (diamagnetic). Some substances, such as iron, are ferromagnetic which cause significant disruption of the local magnetic field. Significant magnetic susceptibility has two

62

Fig. 4.4 Diffusion EPI brain acquisition showing hyperintense signal at the level of the internal auditory canal caused by susceptibility.

| 09.11.19 - 01:53

4.2 Case 2

Fig. 4.5 (a,b) Frequency encoding gradient showing gradient warping due to susceptibility.

Fig. 4.6 Comparison of signal loss and distortion due to the presence of metal in three types of imaging techniques: (a) turbo spin echo, (b) gradient echo, and (c) echo-planar imaging.

alterations induce additional spin dephasing in the transverse plane causing signal loss. Even though the additional field inhomogeneities are mitigated by RF rephasing in spin echo techniques, accelerated T2 decay will still occur. This is not the case for gradient echo techniques, where T2*-related signal loss is the predominant result in high magnetic susceptibility environments (▶ Fig. 4.6). It is important to note that exploiting susceptibility effects to enhance tissue characterization has recently become an active area of research. Susceptibilityweight imaging (SWI) uses the effect of suscepti-

bility to characterize tissue properties and is especially useful for looking at iron content and hemorrhages in the brain.

4.2.4 Resolution Complete elimination of image distortion and signal loss due to large magnetic susceptibility may not be attainable, especially for metal implants or other ferromagnetic substances. However, there are several tactics to reduce the impact on surrounding anatomy. Susceptibility

63

| 09.11.19 - 01:53

Magnetic Resonance Imaging artifacts are most prevalent on gradient echo sequences and high field strengths (3T), so switching (if possible) to spin echo alternatives and lower field strengths (1.5T) are the primary options. Also, one should address the key parameters that affect signal loss and image distortion. The former can best be impacted by using lower times to echo (TEs), while still maintaining the desired image contrast. Distortion can be

64

addressed by increasing the imaging BW, which allows for more encoding frequencies and lessens the degree of spatial misregistration. This also allows for shorter TEs. While other strategies exist (such as reducing voxel size), new MR innovations have been developed to incorporate specific susceptibility-reducing attributes, especially for imaging in the presence of metal implants for orthopedic applications.

| 09.11.19 - 01:53

4.3 Case 3: Appearance of Extra Field-of-view Anatomy

4.3 Case 3: Appearance of Extra Field-of-view Anatomy on the Inferior Portion of Sagittal 3D T2-Weighted Acquisition of the Spine 4.3.1 Background ●

3D MR acquisitions are subject to phase encoding in two directions.

●

Isotropic resolution allows high-resolution reconstructions in secondary directions.

●

Additional image quality issues exist for 3D acquisitions.

4.3.2 Findings ●

Large superior-inferior FOV requires activation of both multichannel head coil and cervical/ thoracic spine coil elements. Lumbar spine coil elements are deactivated.

●

Ghost-like artifact resembling the head appears superimposed on the lower part of the image and on the resultant axial reconstruction.

●

No other abnormal artifacts appear on the upper half of the sagittal image, nor in the left-right direction of the axial series.

●

3D T2 TSE sequence was applied using volume selective RF excitation, but with minimal oversampling of data in the in-plane phase-encoding direction (superior-inferior direction) to save imaging time.

4.3.3 Discussion The example shows clear evidence of anatomical wrap-around artifact, also known as “aliasing,” in the superior-inferior direction with a section of the head superimposed onto the lower thoracic spine. This is also apparent on an axial reformatted slice (▶ Fig. 4.7a). Aliasing occurs in the phase-encoding direction for both 2D and 3D acquisitions when employing Cartesian k-space trajectories. The possibility of aliasing in any imaging scenario depends on three main situations: (1) whether tissue is within the sensitivity region of an activated RF receiver coil; (2) whether the prescribed FOV is smaller than the anatomical extent in the (2D or 3D) phase-encoding direction; and (3) the excitation region of the RF transmission pulse. In ▶ Fig. 4.7, the top of the head is within a region that has both been excited by an RF pulse and observable by the activated head coil. Since the in-plane phase-encode direction is superior-inferior and the FOV is smaller than this “activated” anatomy, the portion of the head outside the FOV is superimposed, or “wrapped,” onto the other side. This is another form of spatial mismapping due to the MR signal encoding process. To understand the phenomena, note that the tissue outside of the prescribed 2D FOV has still been excited by an RF pulse and thus still subject to applied field encoding gradients. The phase encoding process imparts phase shifts between −180 and + 180 degrees within the FOV. However, tissues outside of the FOV still experience phase encoding, but with shifts outside this range (> + 180, or < −180). For example, due to the cyclical nature (i.e., sine wave) of MR signals, a phase shift of + 185 degrees is equivalent to −175 degrees and is “wrapped” onto other encoded steps of −175 degrees as illustrated in ▶ Fig. 4.8. Aliasing can also occur in the slice

Fig. 4.7 (a) 3D sagittal T2, and (b) a corresponding axial reformatted slice of the spine showing faint signs of head anatomy, indicative of signal aliasing.

65

| 09.11.19 - 01:53

Magnetic Resonance Imaging

Fig. 4.8 Graphical representation of phase-encode wrap around. Excited anatomy outside the field of view “acquires” the same phase-encoding step as corresponding tissue within the FOV.

Fig. 4.9 (a, b) Slice encode aliasing shown on axial 3D T2 imaging. Spins excited outside the prescribed imaging volume are encoded similarly as those within the imaging volume. In some cases, 3D aliasing may be mistaken for pathology.

direction in 3D imaging. However, the case in ▶ Fig. 4.7 employed slab-selective RF, so that little to no tissue outside the intended volume (FOV in the slice direction) is excited. Even with slab-selective excitation, some residual aliasing may occur, as shown in ▶ Fig. 4.9, since no selective RF pulses have a perfect slice profile.

4.3.4 Resolution The straightforward remedy for aliasing is to be conscious of the three criteria mentioned above. One must first observe what tissue resides outside the imaging FOV, but within the RF excitation and activated coil sensitivity region. If possible, specific coil elements should be deactivated to avoid encoding this residual signal. However, no receive coils have sharp sensitivity cut-offs. Moreover, the

66

entire coil-sensitivity region may be needed for particular 3D acquisitions. Mostly, users are resigned to “oversample” the tissue outside the FOV, especially if one does not want to increase FOV or coverage for resolution or data limit purposes, respectively. While this strategy is liberally applied in the 2D phase-encode direction, a lesser amount is generally applied in the sliceencoding direction, particularly if slab-selective excitation is used. Phase encode oversampling costs time, but adds SNR, and therefore, should be balanced against other imaging criteria for specific applications. It should be noted, finally, that some aliasing is tolerable: if aliased anatomy does not impinge on the diagnostic region of interest, significant time-savings or optimized resolution can be achieved. Two applications that exploit this are cardiac and phase-contrast imaging.

| 09.11.19 - 01:53

4.4 Case 4

4.4 Case 4: Precontrast, Axial 3D T1-Weighted Gradient Echo with Fat Suppression Shows Adequate Anatomical Detail, but Minor Edge Ripple and Blur that is Presumed to be Motion 4.4.1 Background ●

3D gradient echo imaging of the abdomen is a breath-hold technique that will be subject to motion artifacts if the patient cannot comply.

●

Imaging resolution is 1.4 × 1.7 mm in-plane, with 3 mm (interpolated) slice thickness.

●

The subject appeared cooperative with breath holding instructions, judging from other acquisitions (not shown).

4.4.2 Findings ●

There are several areas of edge ripple and blur in the image, including the liver capsule and portal vein.

●

undersampled in both the phase- and sliceencoding direction to achieve appropriate scan durations. This amounts to a reduced phase resolution and increased slice thickness, respectively. When resolution is too low in a certain direction, truncation (or Gibbs) artifact may occur. Truncation artifact is identified by periodic low- and high-signal intensity ripples emanating from high-contrast edges. While this appearance may closely mimic motion artifact, truncation effects are exclusive to sharp, high-contrast edges, and fade thereafter; there is no replicating of anatomy. This phenomenon is due to the inability of the acquisition to accurately define sharp edges, using the available frequency encoding range, particularly high-frequency data. The consequence is an overestimation and underestimation of the highcontrast structure, which decays with distance (▶ Fig. 4.11). The propagation distance of ripples is longer if more high-frequency data points are absent from the acquisition. In-plane truncation artifacts affect both 2D and 3D acquisitions. However, truncation in the slice direction is unique to 3D imaging. ▶ Fig. 4.12 shows a coronal reconstruction of the original data set, given in ▶ Fig. 4.12. The reconstructed data makes the truncation artifact more

There is also some signal fluctuation in the visceral fat in the retroperitoneum.

●

While most ripples emanate in the anteriorposterior direction, some extend in lateral directions as well.

●

Incomplete breath holding is a source of minor blur in the phase-encode direction; however, the external abdominal wall appears fairly sharp, without indication of signal propagation (▶ Fig. 4.10).

4.4.3 Discussion Given the requirements of breath-hold imaging in the abdomen, 3D acquisitions are significantly

Fig. 4.10 Breath-hold 3D T1 spoiled-gradient echo with fat suppression, showing adequate breath holding, but subtle ringing and edge enhancements (arrows).

Fig. 4.11 Gibbs ripple resulting from a limited frequency-encoding range.

67

| 09.11.19 - 01:53

Magnetic Resonance Imaging conspicuous, especially realizing that the axial 3D volume was acquired with 6 mm (3 mm interpolated) slice thickness. The ripples, which are best seen in the spleen and against the fatsuppressed retroperitoneum, are large due to the low resolution in the slice direction. The significance of this artifact is not appreciated when viewed in the standard axial orientation.

Fig. 4.12 Coronal reconstruction of 3D data set from ▶ Fig. 4.1, showing more prominent truncation artifact due to large slice thickness.

4.4.4 Resolution The 3D truncation artifact seen in ▶ Fig. 4.10 can be improved by reducing the slice thickness, or rather, improving the slice resolution. Similarly, a fully sampled in-plane data set, such as 288 × 288, will also reduce edge ripple in 2D. However, it is important to note that a complete elimination of edge ringing is not possible. This is due to the discrete nature of digitized data sampling in MRI and subsequent Fourier reconstruction. In practice, an infinite frequency range would be necessary to describe a sharp step change in signal intensity between two objects, which is not possible in MRI experiments. However, when high enough resolution imaging is employed, a greater frequency range is available to approximate edge information, with highcontrast features being more well defined (▶ Fig. 4.13). Even though ringing may still be present, both intensity and ripple distance from the edge location will be progressively reduced, which makes them less conspicuous. If increasing the resolution is not practical due to SNR and time constraints, applying a filter to smooth the image helps to reduce much of the ringing, at the expense of some edge blurring. Fig. 4.13 Comparison of low- and high-resolution imaging of a grid phantom. The degree of truncation ripple is less in the high-resolution scan.

68

| 09.11.19 - 01:53

4.5 Case 5: Dark Etching Appears at the Boundary of Fat and Soft-Tissue Layer

4.5 Case 5: Dark Etching Appears at the Boundary of Fat and Soft-Tissue Layers 4.5.1 Background ●

Routine Ax T1 TSE for neck soft tissues with phase encoding left-to-right.

●

Subject is cooperative, and slight blur around the tongue is considered normal.

4.5.2 Findings ●

While some dark signal is attributable to signal loss, specific etching in the locations indicated seems abnormal (▶ Fig. 4.14).

●

Parameters are typical for T1 imaging (TR ~ 500 milliseconds; TE ~ 11 milliseconds), with a matrix of 256 and bandwidth less than

mismapping: when two neighboring tissue proton species have different resonant frequencies, they may be spatially misinterpreted during frequency encoding, and therefore spatially offset from one another (or superimpose) during image formation. This is the case for fat and water protons, where the frequency offset is 3.5 ppm, or 224 Hz at 1.5 T. The degree of water–fat shift (WFS) is inversely proportional to the frequencyencoding bandwidth, as with susceptibility. The prevalence of type 1 artifact to the frequencyencode direction (for non-EPI sequences) is due to the relatively low readout BW of these sequences, as well as the fact that transverse magnetization is either refocused or spoiled for each echo measurement. This effectively negates the accumulation of precession-related offsets in the phase-encode direction. Since EPI utilizes a very high BW, it is a special case where type 1 artifact typically occurs in the phase-encode direction.

200 Hz/pixel.

4.5.4 Resolution 4.5.3 Discussion The appearance of dark etching, also known as “India ink,” between soft tissues and fat is indicative of type 1 chemical shift artifact. Type 1 artifact, which is considered a misregistration artifact between fat and water spins, affects all conventional imaging techniques such as spin echo, TSE, and gradient echo. Chemical shift is somewhat similar to susceptibility-induced

Type 1 artifact does not degrade image quality to the extent of motion or susceptibility artifact. However, it may have impact on diagnosis if water–fat misregistration obscures pathology such as cartilage thickness. As alluded, chemical shift is greatest when the frequency-encode BW is low. But it also becomes more significant when image resolution is also low (large FOV or small matrix). Since the resonant offset between water and fat is well defined (224 Hz at

Fig. 4.14 (a, b) Axial T1 turbo spin echo of the soft tissues of the neck reveal subtle black etching around tissue boundaries (arrows).

69

| 09.11.19 - 01:53

Magnetic Resonance Imaging

Fig. 4.15 Axial T1 turbo spin echo of the neck at two different bandwidths: 200 Hz/px (left), and 425 Hz/px (right). Note that the frequency-encoding direction is left-right in (a) and anterior-posterior in (b). There is noticeable reduction in chemical shift signal loss along boundaries with higher bandwidth.

1.5 T and 448 Hz at 3 T, etc.), one can easily calculate the expected WFS in terms of image pixel offset, if the FOV, matrix, and BW are known (see appendix). Since FOV and matrix are often fixed due to application criteria, increasing BW directly remedies the artifact (▶ Fig. 4.15). This tactic also has less SNR penalty than increases in image resolution, which is another remedy. Alternatively, frequency and phase directions can be swapped, in lieu of any parameter adjustment, as long as other artifacts, such as

70

aliasing and motion, are not adversely affected. One must also be wary of chemical shift at high field strengths, since one-to-one transfer of imaging parameters will not be optimal; a proportional increase in BW is necessary to achieve the same WFS as lower field strength. Another solution to eliminate the appearance of fat shifts is to employ fat saturation. Though effective, this clearly alters the purpose of the sequence, and may not be warranted in particular clinical applications.

| 09.11.19 - 01:53

4.6 Case 6: Application of Fat-Suppressed Sequences

4.6 Case 6: Application of FatSuppressed Sequences in the Pelvis did not Reveal the Expected Contrast 4.6.1 Background ●

Fat-suppressed T1 and T2 sequences are universally employed in body MR applications, such as female pelvic imaging for fibroids.

●

The most common fat saturation technique is chemically selective RF excitation, whereby only fat resonant frequency is specifically targeted with a saturation pulse.

4.6.2 Findings ●

Chemical fat saturation is incomplete in both T1- and T2-weighted imaging (▶ Fig. 4.16).

●

It is likely that the MR system erroneously tuned the center frequency to be the dominant species in the FOV, in this case fat.

●

Alternatively, strong inhomogeneous fields or poor shimming within the FOV may alter the expected off-resonance position of fat species relative to water.

●

Chemical fat saturation RF pulses are automatically applied at 440 Hz (at 3T) down field from the tuned center frequency, which no longer effectively excite fat resonances.

●

Partial or incomplete fat saturation typically appears as regional bands of dark fat tissue.

4.6.3 Discussion Fat suppression has many applications, including making edema and inflammation more conspicuous on T2-weighted imaging and eliminating confounding high signal from postcontrast T1-weighted imaging. One important advantage of a resonant frequency difference between water and fat protons is the ability to selectively saturate the magnetization of fat in MR images. This can be done in a variety of ways, but the most common is to center an additional RF saturation pulse over the resonant frequency of fat. In actuality, fat has up to six different resonant frequencies, with the most significant occurring at 1.3, 2.1, and 0.9 ppm (▶ Fig. 4.17). For this reason, spectrally selective RF pulses must also have a prescribed BW, but must be limited to prevent intruding water resonance at 4.7 ppm. In some areas of the body where complex or abnormal tissue geometry affects the local magnetic field, the prescribed fat-centered RF pulse and BW may partially “miss” the true susceptibility-induced resonant frequency of fat causing incomplete tissue saturation on images. Moreover, this scenario may even lead to erroneous saturation of water signal (▶ Fig. 4.18a). Another root cause for poor or erroneous fat saturation is the inefficiency of RF pulses themselves; sharp frequency cutoffs are difficult to achieve, especially over a small spectral range. Consequently, the bell-shaped profile may cause some varying excitation of resonances inside and outside the frequency bounds. Separate from local susceptibility changes, large FOV imaging also causes regions of poor fat suppression, primarily along the periphery of the FOV.

Fig. 4.16 Fat-suppressed axial (a) T2 single-shot, and (b) T1 3D gradient echo acquisitions did not produce hypointense fat signal as expected. Some regional fat uppression is evident.

71

| 09.11.19 - 01:53

Magnetic Resonance Imaging Fat-containing voxels located along the periphery are significantly far from isocenter, where field inhomogeneity also predominates. This also pertains to multislice axial imaging; poor fat suppression is often seen on first and last slices of axial data sets with large number of slices.

4.6.4 Resolution An immediate solution to poor fat suppression is improving the fat suppression pulses themselves. Using adiabatic RF excitation helps improve the spectrum of targeted fat protons. Alternatively, spectral excitation can be performed on water protons only, whose spectral amplitude and line width are usually more well defined than fat. Another strategy is to convert the acquisition to a short-tau inversion recovery (STIR) technique which offers increased suppression uniformity over broad FOVs and field inhomogeneity (▶ Fig. 4.18b). STIR utilizes a nonselective 180degree inversion (IR) prepulse timed to null the

longitudinal magnetization of fat protons prior to image acquisition, based on the its short T1 value (~ 250 millisecond at 1.5T). Since the IR prepulse affects both fat and water proton resonant frequencies, all tissue will undergo longitudinal T1 recovery. Most tissues relax slower than fat, and will not be suppressed at the selected inversion time (TI); however, they will incur reduced available magnetization, which translates to reduced image SNR. With chemically selective fat suppression methods, it is always useful to observe the spectral peaks of fat and water following any shimming procedure when fat suppression uniformity is desired. Even though broad line widths may still persist, manual frequency adjustments help to resolve significant fat frequency shifts caused by off-resonance. This strategy can be further optimized by using smaller FOVs, or fewer slices, thereby limiting the effective volume of shimming. More systems are now equipped with sophisticated fat-water separation techniques, which Fig. 4.17 Sample spectrum of fat and water resonant peaks from a voxel obtained in fatty liver tissue. Note the broad lipid peak, suggesting the presence of other lipid resonances. Arbitrary units are used along the x-axis.

Fig. 4.18 Comparison of (a) spectrally selective fat suppression, and (b) nonselective inversion-recoverybased fat suppression (STIR) in the cervical spine. Local field homogeneity impacts the uniformity of fat suppression in (a) to the extent that water signal may also be partially affected.

72

| 09.11.19 - 01:53

4.6 Case 6: Application of Fat-Suppressed Sequences

Fig. 4.19 Comparison of (a) spectrally selective fat suppression, and (b) twopoint Dixon water-only images of the neck in two different subjects. While some expected inhomogeneous is expected in (a), more robust suppression is achieved by separating fat and water images with the Dixon technique (b).

evolved from the well-known two-point Dixon method (▶ Fig. 4.19). Modern versions of the method still incur a scan time penalty, but are very efficient for creating robust fat-suppressed (wateronly) images. These techniques do not rely on spectrally selective pulses, or nonselective IR

pulses, which reduce SNR of all tissues. However, the overall efficacy of the reconstruction is highly dependent on producing a suitable domain magnetic field map. Nonetheless, unfavorable fat/ water swapping can persist in regions of significant Bo field inhomogeneity.

73

| 09.11.19 - 01:53

Magnetic Resonance Imaging

4.7 Case 7: T1-Weighted Gradient Echo of the Abdomen Shows Marked Artifact Medially on Both Coronal and Axial FOV, Obscuring Visualization of Soft Tissues 4.7.1 Background ●

Most T1-weighted 2D and 3D acquisitions of the chest, abdomen, and/or pelvis require subject breath hold and some method of scan acceleration.

●

In addition to reducing the phase-encode resolution, utilizing parallel imaging acceleration with multichannel receive coils is another common strategy.

4.7.2 Findings ●

Both examples in ▶ Fig. 4.20 utilize parallel imaging approaches.

●

Relatively small FOV was prescribed in the respective phase-encode directions to further reduce acquisition time.

●

Both images suffer from unwanted signal artifact in relevant soft tissue.

4.7.3 Discussion The availability of parallel imaging methods has tremendously improved the utility of MR in a variety of applications. Parallel imaging involves utilizing multiarray receive coils over the imaged region, and using their individually specific coil sensitivity to reconstruct undersampled k-space data. Scan times can be drastically reduced by 2 × or more with parallel imaging but the side effect is reduced SNR and some associated artifacts. Often significant amplification of noise is seen with parallel imaging when reduction factors exceed 3 × . Furthermore, if multiarray coils are not properly placed around the region of interest, more noise amplification is observed. One parallel imaging method reconstructs undersampled k-space data using multiple coil sensitivity images in image space (e.g., SENSE). Artifacts with this method appear similar to image foldover, although the aliased regions typically appear in the center of the FOV and are sometimes mistaken as signal “hotspots.” Another common parallel imaging technique estimates missing imaging information in k-space (e.g., GRAPPA). If missing k-space data is not effectively recovered, unwanted phase shifts may develop, resulting in ghost-like artifacts in the phaseencode direction (▶ Fig. 4.21).

Fig. 4.20 (a) T1-weighted 3D gradient echo (GRE) of the abdomen shows marked artifact through the middle of the field of view, obscuring visualization of soft tissues. (b) A nonspecific high-signal line also appears on an axial 2D GRE.

74

| 09.11.19 - 01:53

4.7 Case 7

Fig. 4.21 Subtle ghost-like artifact appears in liver across the center of the field of view (arrows). This artifact is not indicative of motion, but rather results from parallel imaging effects.

4.7.4 Resolution Parallel imaging artifacts may mimic physiological or other technical artifacts; therefore, one must first identify the true origin of poor SNR, ghosts, or aliasing. For image-based parallel imaging, some extended phase FOV (oversampling) will alleviate the subtle unfolding reconstruction artifact. Furthermore, effort should be made to match the anatomic positioning between coil calibration scans and pulse sequences using parallel imaging. This may require calibration scans to be performed using similar breath hold instructions.

In k-space-based methods, parallel imaging ghosts are usually rare since autocalibration steps are built into the acquisition. However, the technique is more sensitive to patient motion or inadequate reference coil sensitivities. Often, more autocalibration reference lines are needed, which reduces the scan acceleration. Increased reference lines also alleviate central noise banding, which is common with parallel imaging. Finally, routine system and coil maintenance is vital for optimal performance of sequences using parallel imaging.

75

| 09.11.19 - 01:53

Magnetic Resonance Imaging

4.8 Case 8: Abnormal Dark Fluid Seen in the Bladder of an Axial Single-Shot T2Weighted Sequence, but not on Location-Matched 3D T2 Acquisition

4.8.2 Findings ●

individually in less than a half second, an entire volume of tissue is excited and combined every TR period for the 3D T2 technique. ●

All soft-tissue pelvic imaging involve either large FOV axial T2 for gross anatomy, or small FOV

regions of hypointense signal indicate some flow on HASTE (▶ Fig. 4.22a). ●

slice during the acquisition. This fluid was not

diseases, such as rectal and prostate cancer.

anatomical assessment since it is able to quickly acquire large superior-to-inferior coverage with minimal motion artifacts. ●

HASTE acquires just over half of its k-space data in one TR period, with the rest being interpolated. This enables ultra-fast single-slice acquisition, without T1-weighting.

●

originally excited, but replaces some fluid in this

Often, half-Fourier-acquired single-shot turbo spin echo (HASTE) is performed for overall

High resolution 3D T2 acquires k-space data in multiple TR periods, depending on the amount of total phase-encode data and the acquisition echo train length.

The ultrafast acquisition of HASTE will be sensitive to some flow turbulence entering the

(high resolution) T2 for evaluating specific ●

Static fluid in the bladder will be T2-bright by definition on both sequences, but some focused

4.8.1 Background ●

While each slice in HASTE imaging is acquired

course of time. ●

3D T2 is not sensitive to this subtle fluid motion since its influence is effectively averaged out over the multiple TR periods.

4.8.3 Discussion The case above exemplifies the ever-present concept of flow-related contrast in MRI. In many cases, such as in angiographic applications, flowrelated contrast is expected with sequences especially configured to be sensitive to the phenomena. However, many essential sequences intended for soft tissue analysis may be prone to unexpected

Fig. 4.22 Abnormal dark fluid seen in the bladder of an axial single-shot T2-weighted sequence (a). The appearance is absent from a location-matched 3D T2 acquisition (b).

76

| 09.11.19 - 01:53

4.8 Case 8: Abnormal Dark Fluid Seen in the Bladder hyper- or hypointensities related to flow. It was previously discussed (Case 1) that fluctuating signal intensity due to periodic flow will lead to ghosting if not properly compensated or synchronized with the acquisition. While this is particularly true for TSE sequences, flow-related contrast may also appear as a static indication of a particular flow state and possibly even point to disease abnormalities. In the case of ▶ Fig. 4.22a, the fluid in the bladder is mostly stationary, and predictably appears bright due to its high T2 value. The dark areas represent subtle flow from fluid entering the slice between excitation and TE. In contrast, consider ▶ Fig. 4.23a, b, where blood in the aorta and portal vein are both dark on axial and coronal HASTE, respectively. In rapid flowing vessels, blood that is slice-excited quickly travels out of the slice before the signal is both refocused and measured

(at TE), leaving a “flow void.” As flow becomes restricted or stationary (as in veins or certain parts of the cardiac cycle), blood will appear diffusely hyperintense, similar to fluid, since it also has relatively high T2. This is apparent in ▶ Fig. 4.23c, d, in a patient with metastatic neuroendocrine disease, where HASTE shows some hyperintense blood signal representative of slow or restricted flow. In this case, it is important to examine all slices to determine whether the signal is due to sensitivity to the cardiac cycle or related to disease (such as hypertension or thrombosis). Flow-related contrast also affects gradient echo MR sequences. Since TE and TR are typically shorter than spin echo type sequences, unperturbed blood flowing into the slice is both excited and measured in close tandem, producing a bright blood phenomenon. A wellknown example is 3D time-of-flight (TOF) imaging.

Fig. 4.23 (a) Axial and (b) coronal T2 single-shot in a healthy subject reveals hypointense signal in the aorta and portal vein. Similar acquisitions obtained in a patient with metastatic disease (c and d) show some hyperintense blood signal in the corresponding vessels, suggesting slow or restricted flow.

77

| 09.11.19 - 01:53

Magnetic Resonance Imaging

Fig. 4.24 Examples of entry-flow phenomena on gradient echo (GRE) images. (a) 2D axial GRE, with bright blood in liver vessels; (b) a reconstructed coronal-oblique 3D GRE with hyperintense blood signal, which slowly fades in the direction of flow.

In this sequence, blood entering the imaging volume is excited and measured at TE, but escapes the volume before the next excitation. This “entryflow” enhancement effect can be contrasted with stationary tissue that remains dark due to repeated exposure to RF excitation. In some cases, bright blood on gradient echo is unwanted or unexpected, as shown in ▶ Fig. 4.24, since it may be mistaken for the presence of gadolinium contrast agent.

4.8.4 Resolution As stated, flow-related contrast is desirable in many applications. In addition, it aids in many diagnostic instances. In practice, it is important to understand how flow from generally static fluid or from fat flowing blood may affect the signal properties of specific spin or gradient echo sequences.

78

Lack of flow voids on spin echo type imaging, especially HASTE, are indicative of slow-moving blood remaining in the imaging slice. These diastolic hyperintensities can be lessened with cardiac gating, with or without the addition of a “black blood” pulse. These additional pulses come in a variety of methods (both gradient and RF forms), and help to further eliminate residual blood signal though flow cancellation and dephasing. Conversely, flow voids are unwanted in angiographic 3D gradient echo. It is important to prescribe sufficient volumetric coverage to maximize inflow enhancement, while being conscious that in-plane flow from tortuous vessels may result in some signal loss. For nonangiographic gradient echo, shortened TR and optimized TE, especially for 3D imaging, ensures dark vessel lumen.

| 09.11.19 - 01:53

4.9 Case 9: Postcontrast T1-Weighted Gradient Echo Reveals Patchy Enhancement

4.9 Case 9: Postcontrast T1Weighted Gradient Echo Reveals Patchy Enhancement in the Anterior Septal Wall 4.9.1 Background ●

Corresponding cine and postcontrast T1 is essential for demonstrating areas of scar due to myocardial infarction.

●

Postcontrast T1 is commonly performed with inversion recovery fast gradient echo sequences, which is intended to enhance the contrast between healthy and infarcted myocardium.

●

T1 image contrast is controlled by the time delay between the inversion pulse and data acquisition.

4.9.2 Findings ●

Area of scar enhancement not clearly defined relative to healthy myocardium.

●

Corresponding short-axis cine shows thin myocardium in the same region.

4.9.3 Discussion Delayed contrast enhancement (DCE) is a powerful cardiac MRI technique for distinguishing areas in the myocardium with increased extracellular distribution volume due to cell necrosis, fibrosis, or scarring (▶ Fig. 4.25). This extends

from the pharmacokinetics of most gadolinium contrast agents, which provide significant T1 enhancement in regions of high vascularity and extracellular space. In the case of infarcted myocardium, the scarred region experiences delayed contrast distribution due to poor vascularity followed by prolonged clearance. This is in stark contradiction to the more predictable kinetic behavior in both remote myocardium and blood pool. While fast T1-weighted imaging can exploit the T1 differences between normal and infarcted myocardium, image contrast can be significantly improved by implementing IR preparation pulses. These 180-degree RF pulses, which invert the polarity of all available longitudinal magnetization, are performed at a designated time prior to normal data acquisition. In other words, this process “prepares” the longitudinal magnetization for subsequent data collection. Since all tissues including normal and infarcted myocardium have unique T1 values, IR pulses have the ability to further enhance the dynamic range and contrast if the time of inversion (TI) is selected appropriately. Maximum apparent contrast in the case of delayed myocardial imaging is achieved when TI is selected to cancel (or “null”) the longitudinal recovery of normal myocardium (▶ Fig. 4.26). The TI null point implies tissue signal suppression (in terms of its T1 value), since no longitudinal magnetization exists as imaging data collection commences. ▶ Fig. 4.27 shows another example of how changing the TI value significantly alters the ability to distinguish infarcted from normal myocardium.

Fig. 4.25 (a) Postcontrast T1-weighted gradient echo reveals patchy enhancement in the anterior septal wall. (b) Corresponding short-axis cine shows thin myocardium in the same region.

79

| 09.11.19 - 01:53

Magnetic Resonance Imaging

Fig. 4.26 Plot of the relative longitudinal T1 recovery of contrast-enhanced normal and infarcted myocardium following application of an inversion (180 degrees) RF pulse. The concept is to commence image acquisition after an inversion time (TI) set to suppress the signal of normal myocardium. This occurs at the point when recovery crosses the x-axis.

Fig. 4.27 Comparison of contrast-enhanced short-axis inversion recovery in a subject with subendocardial infarction. A TI of 600 milliseconds (a) does not reveal the extent of infarct region, while adapting to TI = 300 milliseconds (b) shows the enhancement sufficiently (arrow) by virtue of suppressing signal from normal myocardium.

80

| 09.11.19 - 01:53

4.9 Case 9: Postcontrast T1-Weighted Gradient Echo Reveals Patchy Enhancement There are a host of applications that utilize IR. Each carefully prescribes TI to null a particular tissue’s (known) T1, or enhance contrast between two T1 s. Both STIR- and fluid-attenuated inversion recovery (FLAIR)-type sequences employ IR to null fat and fluid, respectively. Some other applications, such as postcontrast brain imaging (e.g., MPRAGE), utilize an optimized TI to enhance grey and white matter differences, in contrast to enhancing lesions. It can be shown that T1 image contrast improves with the use of IR prepulses at the expense of some increased imaging time.

4.9.4 Resolution In general, if T1 is known (e.g., T1fat = 250 millisecond; T1csf = 3500 millisecond), optimal null point TI is calculated from a modified monoexponential recovery equation (TI = 0.693 × T1). An analytical solution can also be made to estimate the expected image contrast of any chosen TI, given multiple T1 values S ¼ 1  2eTI=T1 ▶ Fig. 4.27 shows the important image contrast differences when different TIs are chosen for the

same application. In this case, image contrast between subendocardial infarct and normal myocardium is optimized with TI = 250 millisecond (▶ Fig. 4.27b) compared to TI = 600 millisecond (▶ Fig. 4.27a). However, since contrast agent concentration changes dynamically following administration, simply applying TI = 250 millisecond to the case in ▶ Fig. 4.25 may not always null normal myocardium. Often, a “TIscout” acquisition must be performed, which quickly samples images at multiple TI values. The user then selects the optimal null point TI to subsequently optimize IR delayed enhancement imaging. Similar TI optimization is needed in other sequences, as well, such as FLAIR. In this sequence, CSF fluid suppression is desired. Since CSF fluid has a lengthy T1 (~ 3500 millisecond), which is on the order (or greater than) most spin echo TRs, one must consider the effect of incomplete T1 relaxation when choosing an optimal TI value. In this case, the optimal TI is less than the predicted TI, and scales exponentially with TR. For completeness, it is often important to consider other parameters, such as TR, flip angle, and echo train length, when determining optimal TI, especially for long T1 tissues.

81

| 09.11.19 - 01:53

Magnetic Resonance Imaging

4.10 Case 10: Significant Signal-to-Noise Variation Across the FOV, Creating Nondiagnostic Image Quality 4.10.1 Background ●

MR signal originates from net magnetization aligned along the main magnetic field.

●

Above all, the imaging process requires spin excitation and signal reception by RF coils.

●

The homogeneity of the prescribed transmission and receiving RF fields are proportional to the resulting image SNR.

4.10.2 Findings ●

Significant noise is apparent in the middle of FOV, and anterior to the spinal anatomy (▶ Fig. 4.28).

●

Underlying T2 image contrast is sufficient, despite poor SNR.

●

Brain anatomy is spared from poor image quality.

4.10.3 Discussion Upon closer inspection, one finds that a particular receive coil element surrounding the neck was mistakenly deactivated during the scan. Fortunately, recognizing this error allows a simple remedy to ensure all necessary coils are activated over the region of interest, thus recovering the underlying signal (▶ Fig. 4.29). The persistently lower SNR anterior to spinal anatomy points to the use of primarily posterior receiver coil elements. While this scenario and remedy seems trivial, other more significant root causes may lead to SNR variation across the FOV, such as malfunctioning receiver coils. The common theme, therefore, is to recognize the unequivocal importance of both receive and transmission RF fields in MR image quality. As stated in the background, these RF fields buttress the entire MR image formation process; RF transmission begins the experiment by exciting proton spins, while RF receiver coils (or “antennas”) capture the encoded signal. Loss of

82

integrity in any part of these two elements dramatically reduces the upper threshold of achievable SNR and, hence, image quality. For this reason, it is important to have a routine quality assurance program in place, whereby RF coils are tested for proper functionality. An important distinction must be made when interrogating errors and artifacts related to either RF transmission or RF reception. Particularly, RF transmission field mostly originates from the main magnetic coil located within the bore (even though local transmit/receive and dual transmit coils also exist). This transmission is pretuned based on desired flip angles and emits across a broad volume within the bore. It is clear that the integrity of the RF transmission field is only entirely uniform over a smaller finite volume (~ 30–40 cm) around the isocenter, and falls off thereafter due to inhomogeneity (▶ Fig. 4.30). The result is variation in the expected flip angle distribution, which in turn affects transverse magnetization. Another dramatic effect of RF transmit field inhomogeneity is dielectric effect (▶ Fig. 4.31). In this situation, which mostly occurs at high field strengths (> 1.5T), image signal variation is caused by the RF field’s interaction with various tissue conductivities, particularly increased water content. RF receive coil integrity is equally (if not more) important for optimal SNR MRI. It is entirely possible (and common) that a well-functioning receive coil can produce low SNR images, if care

Fig. 4.28 Significant signal-to-noise variation across the field of view, creating nondiagnostic image quality.

| 09.11.19 - 01:53

4.10 Case 10: Significant Signal-to-Noise Variation Across the FOV,

Fig. 4.29 Comparison of the original acquisition (a), with a second acquisition with the appropriate coils activated (b).

Fig. 4.30 (a) An initial localizer acquisition uses a large field of view (FOV) to routinely locate and identify the anatomy in question. It also reveals the effective region coil sensitivity and signal fall-off. (b) A smaller, more focused FOV is subsequently used for higher-quality anatomic visualization, while being conscious to limit including areas outside the effective sensitivity region.

Fig. 4.31 (a) The increased abdominal fluid affects the RF transmit uniformity. The fluid acts as dampener due to the change transmit conductivity, thereby modifying the effective flip angle. (b) The dielectric effect is absent in patients with minimal to no interabdominal fluid.

is not paid to its proper positioning. This is exemplified in ▶ Fig. 4.32, where there may be legitimate question about the significant drop in SNR across a multislice acquisition, when in

actuality, the coil was placed suboptimally. The skill developed in coil use and placement is often understated and presumed.

83

| 09.11.19 - 01:53

Magnetic Resonance Imaging

Fig. 4.32 (a) Sagittal T2 image of the pelvis does not extend through the pelvic floor due to poor coil coverage in the region. (b) and (c) show corresponding axial acquisitions from two locations. Note the significant drop in SNR in (c) relative to (b) toward the pelvic floor due to poor coil sensitivity.

4.10.4 Resolution Many image quality issues stem from poor coil placement or not recognizing the optimum extent of the RF transmit uniformity. Quality MR would not be possible without careful attention to both items. If an imaging practice intends to perform large FOV exams (e.g., abdomen-pelvis, thoracic spine, long bones, etc.), it is wise to acquire large (> 40 cm FOV) survey acquisitions, and measure the region of optimal signal homogeneity available. This can be appreciated in ▶ Fig. 4.31, where a FOV of 40 cm shows signal drop out beyond a 30 cm radius; subsequent scans are limited to

84

FOV < 30 cm to preserve optimal quality. Initial survey scans also reveal whether receive coils are positioned too high or low (▶ Fig. 4.32). This should be adjusted based on the extent of anatomic coverage. Positioning and activation of coils and coil elements should also play a role when diagnosing other artifacts, such as parallel imaging (Case 7, T1-Weighted Gradient Echo (GRE) of the Abdomen Shows Marked Artifact Medially on Both Coronal and Axial FOV, Obscuring Visualization of Soft Tissues) and aliasing (Case 3, Appearance of Extra Field-ofView Anatomy on the Inferior Portion of Sagittal 3D T2-Weighted Acquisition of the Spine).

| 09.11.19 - 01:53

4.11 Review Questions

4.11 Review Questions 4.11.1 Case 1: Appearance of Discrete Image Ghosts on Abdominal Imaging 1. The acquisition of phase-encode steps at both inspiration and expiration states during the course of free-breathing abdominal MRI will most likely result in what appearance in the

4.11.2 Case 2: A Well-Defined Area of Signal Hyperintensity Appears Bilaterally at the Level of the Internal Auditory Canal on Diffusion-Weighted MRI, Affecting Visualization of Surrounding Structures 4. Which of the following strategies will help

image?

reduce metal susceptibility artifact?

a) Image blurring

a) Increase BW, and switch to fast gradient

b) Discrete image ghosts

echo

c) Ringing in the frequency direction

b) Increase BW, and lower TE

d) Pulsation artifacts

c) Switch to fast gradient echo, and increase resolution and signal averaging

2. Which of the following methods is NOT an effective strategy for reducing motion

d) Swap phase-encode direction, and increase resolution and slice thickness

artifacts? a) Breath holding b) Respiratory gating c) Shallow breathing d) Ultrashort TR sequences 3. “Single-shot” MR abdominal acquisitions, such as half-Fourier turbo spin echo (TSE), collect all

4.11.3 Case 3: Appearance of Extra Field-of-view Anatomy on the Inferior Portion of Sagittal 3D T2-Weighted Acquisition of the Spine 5. To image small anatomy, such as the pituitary,

phase-encode steps from one slice in one TR

without aliasing artifact using an FOV smaller

period, with very short echo spacing. Regarding

than the brain, which of the following tactics

motion artifacts due to free breathing, which of

should NOT be employed?

the following is NOT true about single-shot

a) Sagittal 2D imaging with maximum phase

methods? a) Multislice images may be misregistered with one another b) Subtle ghosts will be apparent along the phase-encode direction c) Image blurring may occur d) Cross-talk and signal saturation may occur between sequentially acquired slices

oversampling b) Axial 3D imaging with nonselective RF pulses and maximum phase oversampling c) Coronal slab-selective 3D imaging some slice oversampling d) Axial 2D imaging using a rectangular FOV in the left-right direction, such that the ears are truncated

85

| 09.11.19 - 01:53

Magnetic Resonance Imaging

4.11.4 Case 4: Precontrast, Axial 3D T1-Weighted Gradient Echo with Fat Suppression Shows Adequate Anatomical Detail, but Minor Edge Ripple and Blur that is Presumed to be Motion

4.11.6 Case 6: Application of Fat-Suppressed Sequences in the Pelvis did not Reveal the Expected Contrast

6. In 3D imaging, how can one better distinguish

a) Repositioning the subject at the magnet

truncation from motion artifact?

10. All of the following strategies will help to improve fat suppression, except: isocenter

a) Reconstruct data in another dimension

b) Switching to STIR acquisitions

b) If ringing occurs in all directions, it is due to

c) Switching receiver coils

motion c) Signal intensity is much higher with

d) Shimming and manually adjusting the center frequency

truncation d) There is no way to distinguish them

11. Frequency-selective fat saturation is usually most uniform in which of the following

7. Truncation artifact may occur in the frequencyencode direction when: a) Resolution is very high b) BW is large c) There are very high-contrast boundaries along that direction d) A smoothing filter is applied

scenarios? a) Anatomy that allows small FOV, such as the knee b) Anatomy with irregular geometry, such as the neck c) Off-center FOV, such as the elbow at one’s side d) Large FOV, such as bilateral femurs and hips

4.11.5 Case 5: Dark Etching Appears at the Boundary of Fat and Soft-Tissue Layers 8. A water-fat shift (WFS) of two pixels at 1.5 T implies: a) Water and fat are off-resonant by 2 times 224 Hz (448 Hz) b) Water and fat will be misregistered by two pixels in the phase-encode direction c) The BW in Hz/pixel must be half of 224 Hz (112 Hz) d) FOV must be too large

4.11.7 Case 7: T1-Weighted Gradient Echo of the Abdomen Shows Marked Artifact Medially on Both Coronal and Axial FOV, Obscuring Visualization of Soft Tissues 12. Given that SNR decreases with the use of parallel imaging, what can be done to improve relative SNR and lessen parallel imaging artifacts? a) Oversample in the phase-encode direction b) Reduce resolution c) Increase slice thickness

9. Which imaging scenario would reduce the

d) Use more slices

conspicuity of type 1 chemical shift? a) Removing fat sat from EPI b) Swapping phase- and frequency-encode directions c) Switching from spin echo to gradient echo sequences d) Increasing the imaging resolution

86

13. In spine MRI, typically only posterior receive coils are used. In terms of using parallel imaging, this implies: a) Parallel imaging can be used in any direction b) For sagittal imaging, it is best to use parallel imaging in the superior-inferior direction

| 09.11.19 - 01:53

4.11 Review Questions c) For axial imaging, parallel imaging in the anterior-posterior direction is preferred d) Parallel imaging should not be used

b) SNR is too low c) The short TI may inadvertently suppress contrast agent effects d) Too many flow artifacts

4.11.8 Case 8: Abnormal Dark Fluid Seen in the Bladder of an Axial Single-Shot T2-Weighted Sequence, but not on LocationMatched 3D T2 Acquisition 14. Low signal intensity of cerebrospinal fluid (CSF) on T2-weighted cervical spine imaging is likely due to: a) Flow ghosting b) Use of saturation bands c) Spin dephasing due to pulsatile flow d) Flow compensation 15. T2-weighted TSE sequences are more sensitive to flow voids than T1-weighted TSE because: a) T2 values of blood are short

17. The null point TI is modified from 225 to 300 millisecond during two DCE scans 5 minutes apart. This implies: a) Contrast agent is mostly in the blood pool b) Fat tissue needs to be suppressed for optimal contrast c) TR is also short d) Contrast agent is slowly being cleared from normal myocardium

4.11.10 Case 10: Significant Signal-to-Noise Variation Across the FOV, Creating Nondiagnostic Image Quality 18. Another possible cause for signal loss in the following image is:

b) TE is higher, giving more time for flowing protons to “escape” the imaging slice c) The multitude of 180 refocusing pulses associated with T2 TSE continually saturates flow d) T1 TSE sequences allow more complete recovery of flow signal each TR period

4.11.9 Case 9: Postcontrast T1Weighted Gradient Echo Reveals Patchy Enhancement in the Anterior Septal Wall 16. STIR acquisitions are typically not used for postcontrast imaging because: a) Spectrally selective fat saturation is good enough

87

| 09.11.19 - 01:53

Magnetic Resonance Imaging a) Cardiac motion and flow b) Metal artifact

Glossary B0: main magnetic field strength

c) A long MR bore d) Having too many coil elements activated 19. One remedy for dielectric effect seen in the

TR: repetition time TE: time to echo Ny: pixels in the phase (y) direction Nx: pixels in the frequency (x) direction

following image is:

Nz: pixels in the slice (z) direction NSA: number of signal averages Δx: resolution (mm) in the frequency direction Δy: resolution (mm) in the phase direction Δz: resolution (or slice thickness) in the slice direction BWread: bandwidth in the frequency readout direction WFS: water–fat shift (in pixels) FOV: field-of-view Δf: relative frequency offset between two proton types (e.g., fat and water) Δχ: susceptibility difference between two tissues g: geometric factor related to coil configuration and placement R: parallel imaging acceleration factor a) Position the patient prone b) Omit T2 imaging from the protocol

if possible

Bernstein MA. Signal acquisition and k-space sampling. In: Bernstein MA, 1st ed. Handbook of MRI Pulse Sequences. Burlington, US: Elsevier Academic Press; 2004:367–442

Equations

Brown MA, Semelka RC. MRI: Basic Principles and Application. 2nd ed. New York, NY: John Wiley & Sons, Inc; 1999

c) Reposition the receive coil d) Attempt the exam at lower field strength,

Scan time ¼ TR  Ny  Nz  NSA sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi! NSA  Nx  Ny  Nz SNR / K ðΔx  Δy  ΔzÞ BWread WFS ¼

FOV  Δf Nx  BW

Susceptibility artifact size e ðΔxÞ  TE  B0 =BWread SNR e

88

Suggested Reading

1 pffiffiffi R

g

Brown RW, Cheng YC, Haacke EM, Thompson MR, Venkatesan R. Magnetic Resonance Imaging: Physical Principles and Sequence Design. 2nd ed. Hoboken: Wiley; 2014:944 Graves MJ, Mitchell DG. Body MRI artifacts in clinical practice: a physicist’s and radiologist’s perspective. J Magn Reson Imaging. 2013; 38(2): 269–287 Griswold MA, Jakob PM, Heidemann RM, et al. Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magn Reson Med. 2002; 47 (6):1202–1210

| 09.11.19 - 01:53

4.11 Review Questions Hargreaves BA, Worters PW, Pauly KB, Pauly JM, Koch KM, Gold GE. Metal-induced artifacts in MRI. AJR Am J Roentgenol. 2011; 197(3):547–555

loading in the brains of patients with β-thalassemia major. AJNR Am J Neuroradiol. 2014; 35(6): 1085–1090

Hamilton J, Franson D, Seiberlich N. Recent advances in parallel imaging for MRI. Prog Nucl Magn Reson Spectrosc. 2017; 101:71–95

Reeder SB, Atalar E, Bolster BD, Jr, McVeigh ER. Quantification and reduction of ghosting artifacts in interleaved echo-planar imaging. Magn Reson Med. 1997; 38(3):429–439

Huang SY, Seethamraju RT, Patel P, Hahn PF, Kirsch JE, Guimaraes AR. Body MR imaging: artifacts, kspace, and solutions. Radiographics. 2015; 35(5): 1439–1460 Lee VS. Cardiovascular MRI: Physical Principles to Practical Protocols. Philadelphia: Lippincott Williams & Wilkins, 2006 Mugler JP, III. Optimized three-dimensional fastspin-echo MRI. J Magn Reson Imaging. 2014; 39 (4):745–767 Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magn Reson Med. 1999; 42(5):952–962 Qiu D, Chan GC, Chu J, et al. MR quantitative susceptibility imaging for the evaluation of iron

Simonetti OP, Kim RJ, Fieno DS, et al. An improved MR imaging technique for the visualization of myocardial infarction. Radiology. 2001; 218(1): 215 Wang J, He L, Zheng H, Lu ZL. Optimizing the magnetization-prepared rapid gradient-echo (MPRAGE) sequence. PLoS One. 2014; 9(5):e96899 Wheaton AJ, Miyazaki M. Non-contrast enhanced MR angiography: physical principles. J Magn Reson Imaging. 2012; 36(2):286–304 Zaitsev M, Maclaren J, Herbst M. Motion artifacts in MRI: a complex problem with many partial solutions. J Magn Reson Imaging. 2015; 42(4): 887–901

89

| 09.11.19 - 01:53

5 Nuclear Medicine Jonathon A. Nye, James R. Galt, and John N. Aarsvold

Introduction Nuclear medicine imaging comprises planar projection and tomographic image acquisition techniques to capture the distribution of radiolabeled substances. Acquisitions includes single time point imaging, gated and time-series imaging to map radiotracer biodistributions for evaluation of normal/abnormal function. Data collection is a photon starved process, that is, the total number of photons collected is substantially lower than any other X-ray-based modality leading to images of high noise. Moreover, attenuation and scatter within the patient and detector coupled with an imperfect photon detection process (detector energy resolution, collimation, and event processing) degrade the image contrast and spatial resolution. Evaluation of these factors, and improvements when possible, is key to the pursuit of improved image quality. Elements of good image quality in nuclear medicine include absence of image distortion and a noise distribution that is consistent with the expected radiotracer uptake.

90

Common Image Quality Problems ●

Patient motion: degrades spatial resolution due to object motion blurring.

●

Improper attenuation correction: leads to hypo- or hyperintense regions due to misalignment between the transmission and emission data in the reconstruction process.

●

High-Z materials: oral/intravenous contrast and implanted metallic devices are opaque to X-rays used in positron emission tomography/computed tomography (PET/CT) systems to collect transmission data leading to artifacts that propagate to the emission image.

●

Truncation: tissue lying outside the camera field of view (FOV) that is not incorporated properly into the reconstruction process.

| 09.11.19 - 01:53

5.1 Case 1

5.1 Case 1: Degraded Resolution of a Whole-Body Planar 99mTc Methylene Diphosphonate Image 5.1.1 Background ●

Patient suspected of having neoplastic disease.

●

Delayed phase skeletal images obtained approximately 2 hours post injection.

●

Anterior/posterior planar contiguous bone imaging protocol with low-energy highresolution (LEHR) collimators.

●

Counts from the first images starting at the head totaled approximately 2.3 million.

5.1.2 Findings ●

Counts are sufficient in each view (▶ Fig. 5.1).

●

A large visual difference in image resolution is observed between the anterior and posterior planar images.

●

The anterior image is of unacceptable diagnostic quality.

5.1.3 Discussion Review of the acquisition setup showed that the anterior camera head was positioned further from the patient than the posterior camera head. The consequence was a sharp degradation in image quality observed in the anterior camera due to the strong dependence of resolution on the collimator Fig. 5.1 Anterior (left) and posterior (right) whole-body 99mTc methylene diphosphonate images.

91

| 09.11.19 - 01:53

distance from the source. The loss of resolution with collimator distance is a linear relationship with distance from the camera as shown in ▶ Fig. 5.2. Although the distance has a large effect on resolution, count rate is not strongly dependent on the source distance from the collimator. The counts collected at a single point decrease by 1/(source-to-collimator distance)2; however, the number of collimator holes that permit passage of photons is proportional to (source-tocollimator distance)2.1-3 Therefore, the total number of counts, represented by the area under the curves in ▶ Fig. 5.2 is essentially the same at all source distances with a parallel-hole collimator. This relationship applies to point, line, and uniformly distributed sources.

Source-to-collimator distance

Nuclear Medicine

5.1.4 Resolution Proper placement of the gamma camera heads relative to the patient is the responsibility of the camera operator. There are no software corrections to sharpen a blurred image, therefore the operator should rescan the patient with the gamma camera heads placed at the proper distances from the patient. A satisfactory bone scan is shown in ▶ Fig. 5.3 with the anterior camera closer to the patient.

92

Fig. 5.2 Resolution of a point source (e.g., point spread function) versus distance from the face of a parallelhole collimator. The total counts are the same for all distributions.

| 09.11.19 - 01:53

5.1 Case 1

Fig. 5.3 99mTc methylene diphosphonate anterior and posterior images after adjustment of camera head position.

93

| 09.11.19 - 01:53

Nuclear Medicine

5.2 Case 2: Effect of Positron Range on Image Quality and Resolution 5.2.1 Background ●

A rubidium-82 (82Rb) chloride PET study is performed to assess myocardial perfusion.

●

Following the resting exam, an fluorodeoxyglucose ([18F]FDG) viability study is performed to assess myocardial metabolism.

●

Both studies are performed in the resting state on the same scanner with the same reconstruction parameters.

5.2.2 Findings ●

The resting [82Rb]CI and [18F]FDG viability studies are of excellent quality exhibited by the high contrast (> 2:1) between the myocardium and blood pool.

●

There is a marked difference in resolution between the [18F]FDG and [82Rb]CI images, where the [82Rb]CI image appears to be blurred relative to the [18F]FDG study.

5.2.3 Discussion A resting/viability protocol is a good example of how positron range affects image quality. Because both images were collected on the same instrument using the same reconstruction

94

parameters, the difference in image quality can be attributed entirely to the position range effect (▶ Fig. 5.4). Note that some differences in image contrast exist due to the biochemical processes that govern the uptake and distribution of [82Rb] CI and [18F]FDG. ▶ Fig. 5.5 details three main factors that affect image resolution in PET imaging: detector size, noncollinearity of the annihilation photons and position range.4,5 Briefly, resolution is approximately proportional to half the detector size for annihilation events originating in the center of the FOV and approximately equal to the detector size at the periphery. Noncollinearity in the annihilation photons occurs because of a small amount of residual momentum remaining at the time of positron annihilation. Since the PET system assumes collinear photons, the line along which the system assigned the coincidence event is in error compared to the true annihilation location. This error in positioning increases with FOV diameter. Lastly, the energy of positrons from beta decay have a continuous spectrum from zero to a maximum energy. As a result of this energy distribution, the range of a position in tissue can be described by an exponential function. The higher the maximum energy, the larger the positron range and distance of the annihilation event from the decay origin. This disparity between location of the radiotracer and annihilation event contributes to resolution loss independently of the detector design. Rb-82 has maximum positron energy of 3,400 keV compared to F-18 at 635 keV. Therefore, all things the same, the resolution achieved with Rb-82 will be inferior to that of F18-labeled compounds.

| 09.11.19 - 01:53

5.2 Case 2: Effect of Positron Range on Image Quality and Resolution

Fig. 5.4 Resting [82Rb]RbCl and [18F]FDG positron emission tomography images oriented along the short, vertical, and horizontal axis.

95

| 09.11.19 - 01:53

Nuclear Medicine

Fig. 5.5 Illustration of resolution degrading factors is the positron emission tomography detection process. The three main factors are noncollinearity of the annihilation photons, detector size, and position range.

5.2.4 Resolution The resolution difference between [82Rb]CI and [18F]FDG in this example is normal and attributed to positron range. Newer reconstruction

96

algorithms now model the positron physics (among other resolution degrading factors) in the iterative reconstruction process, which can compensate for some of this effect.

| 09.11.19 - 02:54

5.3 Case 3: Standardized Uptake Value in Positron Emission Tomography

5.3 Case 3: Standardized Uptake Value in Positron Emission Tomography (Noise Bias) 5.3.1 Background ●

●

Although the ROIs are drawn in the same location for each reconstruction, extracted information differs.

●

▶ Fig. 5.6 reports the SUV mean and max results.

reconstructions are presented.

5.3.3 Discussion

Regions of interest (ROIs) are drawn in the liver,

The SUV is calculated by normalizing the activity concentration in the reconstructed image to the ratio of the administered activity and body weight.6,7 h i Bq image mL SUV ¼ body weight ½g  administered activity ½Bq

Standardized uptake value (SUV) mean of the liver is a semiquantitative indicator of scanner calibration and may be checked periodically to ensure correct SUV calculation.

●

●

Three difference whole-body [18F]FDG PET

which is assumed to be uniform. ●

5.3.2 Findings

The three images are reconstructed with the same iterative algorithm but with bed times of 30, 60, and 90 seconds.

It has meaning similar to the pharmacological concept of distribution volume. That is, if the administered activity is uniformly distributed throughout the body, the SUV would be 1.0 g/mL

Fig. 5.6 A whole-body [18F]FDG coronal slices reconstructed with imaging durations of 30, 60, and 90 seconds per bed position. The max and mean standardized update value are reported for a region of interest placed in the liver (red circles).

97

| 09.11.19 - 01:53

Nuclear Medicine everywhere. This concept is independent of the amount of FDG administered and is a method for removing variation between patients of [18F]FDG distribution that are related to administered amount and body weight. Assessment of the mean SUV in a large uniform area, such as the liver, can be a good marker for detecting potential imaging problems. In uniform normal liver, the mean SUV has been shown to vary with camera manufacturer but have an expected value ranging between 1.8 and 2.3 g/mL. Clinically, the maximum SUV, not the mean, is used in the reporting and staging of cancer with [18F]FDG. SUVmax is the brightest (or highest) voxel in the ROI and has been shown to be a better predictor of outcomes. A number of factors affect the accuracy of SUV including presence of body fat, patient diet/fasting, reconstruction parameters, scan duration, partial volume effects, and others. In this case, we illustrate the change in SUV on scan duration by reducing the number of counts used in each reconstruction while

98

maintaining all other variables constant. As the scan duration per bed is reduced, the noise in the reconstructed images increases. Noise is a highfrequency component in the image; therefore, SUVmax will always increase when noise increases. This can be thought of as a noise bias, where the mean value is less sensitive. Although the SUV mean is robust in the liver, it is less reproducible than SUV max in smaller features such as lesions.

5.3.4 Resolution Image quality is largely subjective and based on the preference of the interpreting radiologist or nuclear medicine physician. It is critical that the patient preparation and PET camera used to collect data be kept the same for each scan in a patient's cancer assessment. Therefore, any bias related to the instrument hardware or reconstruction is largely kept constant throughout a patient’s initial staging and clinical follow-up.

| 09.11.19 - 01:53

5.4 Case 4: The Impact of Attenuation Correction in PET

5.4 Case 4: The Impact of Attenuation Correction in PET 5.4.1 Background ●

An [18F]FDG PET/CT image is collected as part of

●

The overall image is unremarkable and consid-

cancer staging. ered of good technical quality.

5.4.2 Findings ●

A photopenic area is observed at the interface between the lung and liver in the attenuationcorrected [18F]FDG PET (▶ Fig. 5.7a).

●

The height of the liver in the PET scan does not match to that in the CT scan (▶ Fig. 5.7a–c).

●

The nonattenuation-corrected (NAC) scan shows liver activity in areas that the attenuationcorrected scan does not, suggesting these photopenic areas are not physiological (▶ Fig. 5.7d).

5.4.3 Discussion The photopenic area observed in ▶ Fig. 5.7a is not present in the NAC images in ▶ Fig. 5.4b. Also, the

height of the liver in the attenuation-corrected images and CT are the same suggesting that the attenuation correction was not accurate in that interface region. This attenuation-related artifact is common in whole-body [18F]FDG PET and is caused by a difference in temporal resolution between the acquisition of the PET and CT data.8 The result is that the PET represents a time-averaged position of all structures that move during the respiratory cycle while the CT captures a snapshot at a specific phase of the respiratory cycle. This leads to a mismatch in position of structures within the thoracic cavity, more easily observed where the motion extent is the greatest, near the lung and liver boundary. Attenuation correction is the largest data correction step in the image reconstruction process. For a set of annihilation photons that originate at center of the body and travel to opposing detectors in the coronal plane of ▶ Fig. 5.4, the correction factor can be as large as 18 × for a total of 30 cm tissue. If the amount of tissue traversed by these photons were underrepresented by 40%, such as the presence of lung instead of liver tissues, then attenuation correction factor applied would be 5.5 or 3 times lower. An undercorrection in attenuation of this magnitude will alter image contrast leading to photopenic regions similar that described in ▶ Fig. 5.7.

Fig. 5.7 (a) Attenuation corrected coronal [18F]FDG positron emission tomography (FDG PET) slices showing a photopenic area at the interface between the liver and lungs. (b) Nonattenuation-corrected coronal slice with no photopenic regions. (c) Corresponding coronal CT used for attenuation correction of (a). (d) Attenuationcorrected coronal FDG PET using an average CT protocol that better accounts for the liver position in the PET data.

99

| 09.11.19 - 01:53

Nuclear Medicine

5.4.4 Resolution Methods have been proposed to correct for the motion of the diaphragm in attenuation correction but these techniques usually add to the CT radiation dose and are not widely available in clinical

100

practice. Attenuation-related artifacts also lead to scatter correction problems because this also relies on an accurate CT. The common approach is to “read around” the artifact but take caution if a lesion were present in the photopenic area.

| 09.11.19 - 01:53

5.5 Case 5: Iterative Reconstruction and Choosing the Number of Iterations

5.5 Case 5: Iterative Reconstruction and Choosing the Number of Iterations and Subsets 5.5.1 Background ●

A new single-photon emission computed tomography (SPECT) system is being installed and this camera will perform, among other studies, 99mTc hexamethylpropyleneamine oxime (HMPAO) studies for assessment of cerebral blood flow.

●

The imaging center has collected its first patient and is evaluating the iterative reconstruction parameters to determine suitable iterations and subsets.

5.5.2 Findings ●

A series of image reconstructions were made with varying iterations and subsets (▶ Fig. 5.8).

●

Attenuation correction was performed with the Chang method and mu-value of 0.125/cm.

●

is repeated several times until the simulated guess image best approximates the measured projection data. Each pass through the scan simulation process and update of the guess is called an iteration. Note that each iteration increases the resolution and contrast of the final image but also the noise. The primary advantage of iterative reconstruction is the ability to model aspects of the scanning process, such as attenuation, scatter, collimation, and other features that affect the image appearance. A disadvantage of iterative reconstruction is the lengthy computational time needed to run these simulations. The iterative reconstruction process can be sped up by dividing the projection space into subsets, for example, simulation of 16 projections (1 of 8 subsets) out of a 128-projection acquisition. Although only a few projections are simulated, the scan simulation process occurs much faster and the entire image is updated. The result is that the guess converges to the measure projection data faster than simulation of the entire projection space at once. Two iterations with six subsets result in roughly the same image contrast as six iterations with two subsets, and this is computationally faster (▶ Fig. 5.9).

A Butterworth smoothing kernel was applied with a power 10 and cutoff of 0.7 cycles/mm.

5.5.3 Discussion Iterative reconstruction begins with a guess image, typically a uniform image as not to bias the result, followed by a simulation of the scanning process. The scan simulation process includes a comparison in projection space of the guess data with the measured projection data. A correction image is made from this comparison and used to update the guess image. This process

5.5.4 Resolution Iterative reconstruction provides many advantages such as modeling the physical properties of the acquisition system that can lead to more accurate images.9 Each iteration drives the guess image closer to convergence but at the cost of increasing noise. It is common to add a filter to control for the increase in noise. The most appropriate clinical reconstruction will depend on the physician preferences and the exercise shown in ▶ Fig. 5.5a is commonly done to make those decisions.

101

| 09.11.19 - 01:53

Nuclear Medicine

Fig. 5.8 99mTc hexamethylpropyleneamine oxime images created by varying the number of iterations and subsets. Note that the resulting contrast is similar when the iterations × subset product is the same (images along the diagonal). At high iterations and subsets, the Butterworth filter dominates the image noise and contrast.

102

| 09.11.19 - 01:53

5.5 Case 5: Iterative Reconstruction and Choosing the Number of Iterations

Fig. 5.9 Illustration of an ordered subset iterative reconstruction process. The initial guess image is projected by the scan simulation process. The simulated guess projections are then compared to the measured projections. A correction image is created and used to update the guess. The process is repeated (an iteration) until the simulated scan projections of the guess are a good approximation of the measured projections.

103

| 09.11.19 - 01:53

Nuclear Medicine

5.6 Case 6: The Effects of Image Smoothing 5.6.1 Background ●

A myocardial perfusion imaging study was performed with 99mTc sestamibi.

●

The reconstructed transaxial images show a high amount of noise than typically observed (▶ Fig. 5.10).

5.6.2 Findings ●

The counts per projection and the number of projection samples are consistent with published guidelines.

●

Review of the reconstruction protocol shows that iterative reconstruction was performed with 8 iterations and 10 subsets (▶ Fig. 5.11).

●

Images were not corrected for attenuation.

Fig. 5.10

99mTc

5.6.3 Discussion All nuclear medicine imaging systems employ image smoothing to improve contrast through the reduction of statistical noise or to enhance edges to detect boundaries. Filtering can be performed in either the frequency domain, commonly integrated into the reconstruction process, or the spatial domain post reconstruction. An image can be described in terms of summation of different frequencies, where highfrequency components contain edge information (e.g., air–tissue boundary) and low-frequency components describe slow varying features within structures. The application of a filter either removes or modifies the frequency components that are a part of an image, thereby changing its appearance. A common filter in general nuclear medicine is the Butterworth filter, applied in frequency space, used to remove high-frequency components but preserve low-frequency components. Two parame-

sestimibi myocardial perfusion images in the transaxial plane.

Fig. 5.11 A representative transaxial from a 99mTc sestamibi myocardial perfusion exam processed without filtration and with Butterworth filters of 0.4/cm cutoff and power of 2, 5, and 20.

104

| 09.11.19 - 01:53

5.6 Case 6: The Effects of Image Smoothing ters specify the filter behavior: the cutoff frequency and power. The maximum cutoff frequency is 0.5 pixels/cycle (e.g., Nyquist criterion) and lowering the cutoff frequency will remove highfrequency components improving low-contrast features. The power describes how fast the filter reaches the cutoff value and higher power preserves more edge information.10

5.6.4 Resolution The addition of a Butterworth filter to remove highfrequency components from the image substantially improves the image contrast. The trade-off is loss of edge information and blurring of boundaries but, if not overdone, this is acceptable given the improvement in low-contrast resolution.

105

| 09.11.19 - 01:53

Nuclear Medicine

5.7 Case 7: Choosing the Correct Acquisition Image Matrix Size 5.7.1 Background ●

Three-hour delayed 99mTc methylene diphosphonate (MDP) SPECT data were acquired to better evaluate location and extend of disease.

●

Single anterior-posterior projection and sagittal/ transaxial SPECT are evaluated.

5.7.2 Findings ●

The images are of poorer diagnostic quality than typical (▶ Fig. 5.12).

●

The administered activity and counts per projection were normal.

●

The acquisition settings were: low-energy highresolution (LEHR) collimator, 180-degree orbit, elliptical orbit, 120 stops of 20 second each, 64 × 64 acquired projection matrix.

●

The SPECT data was reconstructed with an ordered-subset expectation maximization (OSEM) algorithm with 6 iterations and 16 subsets that incorporates modeling of the collimator resolution.

5.7.3 Discussion A number of variables can cause degraded resolution of the reconstructed images including the choice of collimator, camera positioning relative to the patient, and reconstruction parameters.1 For a typical SPECT system, the expected resolution with a LEHR collimator is approximately 6 to 8 mm at a distance of 10 cm from the collimator surface. Of critical importance is to appropriately sample the imaging space in order to utilize the capable resolution of the camera. The planar projections used to reconstruct a SPECT image consist of discrete samples (pixels) covering the useful FOV. Choice of acquired projection matrix will affect the resolution of the downstream processes, such as SPECT reconstruction. The rule of thumb is to set the sampling distance (pixel size) to one-third of the camera resolution. For example, a camera with an extrinsic resolution of 8 mm and square FOV of 350 × 350 mm should have 131 samples (pixels) in each direction. SPECT instruments offer matrix sizes scaled in powers of 2 (64, 128, 256, 1024) and choosing 128 × 128 would provide adequate sampling in this example. It should be noted that angular sampling must also be set correctly and can be estimated from relationships given in the references. In ▶ Fig. 5.12, the FOV is 300 mm and a matrix size to 64 × 64 is not enough pixels to satisfy the

Fig. 5.12 Anterior-posterior projection and single-photon emission computed tomography sagittal and transaxial slices from a 99mTc methylene diphosphonate bone study.

106

| 09.11.19 - 01:53

5.7 Case 7: Choosing the Correct Acquisition Image Matrix Size sampling rule. This undersampling in the projection space leads to loss of resolution observed in ▶ Fig. 5.13 when comparing the 64 and 128 matrix projections and SPECT reconstructions. Note that the noise is increased in the 128 × 128 matrix as there are fewer counts per pixel for the same total counts. Increasing the matrix size to 256 × 256 or higher makes no improvements on the system’s ability to resolve structures and further increases noise.

5.7.4 Resolution Increasing the matrix size of the planar projections from 64 × 64 to 128 × 128 fully utilizes the available system resolution resulting in improved spatial resolution and image quality. Applying a Butterworth filter with a high power may improve the noise texture of data collected with the 128 × 128 matrix with limited loss is spatial resolution.

Fig. 5.13 Anterior-posterior projection and single-photon emission computed tomography slices in the sagittal and transaxial plan from projection data acquired with a 64 × 64 matrix (top) and 128 × 128 matrix (bottom). Each acquisition has the same total counts.

107

| 09.11.19 - 01:53

Nuclear Medicine

5.8 Case 8: Assessing Patient Motion in Myocardial Perfusion Imaging 5.8.1 Background ●

Myocardial perfusion imaging with 99mTc sestamibi.

●

The short axis images are of poor in quality (▶ Fig. 5.14).

5.8.2 Findings ●

There is some distortion in the short axis heart size, particularly along the septal and inferior wall.

●

Inspection of the planar images shows axial motion in the table direction in some but not all projections (▶ Fig. 5.15).

5.8.3 Discussion The lengthy imaging time of myocardial perfusion SPECT makes this data acquisition process particularly susceptible to patient motion.11 Patient motion such as contractile cardiac, respiratory action or movement on the table can degrade image resolution and introduce artifacts that often appear as structural distortions. Cardiac and respiratory action can be addressed using gating techniques but patient body motion on the table cannot be addressed using these tools. Patient motion can be minimized through

good technologist instructions or using light restraints. The step-wise collection of projection data provides another means to assess motion by playing them back in a loop to view changes in the heart’s position. Up and down shifts indicate motion along the superior/inferior direction (e.g., table direction), whereas lateral/medial shifts are more difficult to assess due the angular sampling of the SPECT acquisition. Tools are available from several manufacturers to minimize patient motion artifacts. These tools permit manual (or automatic) shifts of the projection data. For example, in ▶ Fig. 5.15, shifting projections 17 and 23 approximately two pixels would correctly align the heart with the neighboring projections. ▶ Fig. 5.16 shows the same data in ▶ Fig. 5.14 after the projection data was motion corrected for the superior/inferior movement. Note the improved image quality and remediation of structural distortion compared to the motion-corrupted reconstruction. A note of caution is that shifting projections can do more harm than good and automated routines should be reviewed for accuracy before sent to reconstruction.

5.8.4 Resolution Myocardial perfusion imaging performed with SPECT should be examined for possible motion corruption. This can be easily assessed by viewing playback of the planar projection data for changes in the heart position. Tools that correct for voluntary motion work well but should be reviewed for accuracy if used in an automated state.

Fig. 5.14 Reoriented short axis slices of a 99mTc sestamibi myocardial perfusion study. Note the distortion in the superior and septal walls.

108

| 09.11.19 - 02:56

5.8 Case 8: Assessing Patient Motion in Myocardial Perfusion Imaging

Fig. 5.15 A sample of planar projections used in the reconstruction of ▶ Fig. 5.8 (original panel). The red arrows show a dip in the heart position indicating the patient moved down in the direction of the table. Following motion correction (motion corrected panel), the heart is aligned in all projections.

Fig. 5.16 Reoriented reconstructed short axis slices of a correction of the planar projections.

99mTc

sestamibi myocardial perfusion study following motion

109

| 09.11.19 - 01:53

Nuclear Medicine

5.9 Case 9: Bremsstrahlung Imaging of 90Y Microspheres Liver Embolization 5.9.1 Background ●

A patient workup for intra-arterial resin microsphere radioembolization of a hepatic lesion.

●

They have received a 99mTc microaggregated albumin (MAA) pretreatment planning scan.

●

Post radioembolization SPECT/CT data collection of 90Y bremsstrahlung photons was performed.

●

Data are fused with contrast enhance MRI to confirm the distribution of trapped microspheres is in the tumor vasculature.

5.9.2 Findings ●

Setup of the 90Y SPECT camera includes use of a medium-energy collimator.

●

There is no photopeak to window, as with 99mTc imaging. A 35-keV energy window centered on 108 keV was selected.

●

interactions in soft tissue as it slows down resulting in a wide spectrum of photons with a maximum energy of approximately 2.3 MeV. Since there is no distinct photopeak, as typical for gamma camera imaging, placement of an energy window is not trivial (▶ Fig. 5.17). The placement and width of the imaging window have been suggested to be around 108 keV with a 35 keV width when using a medium-energy collimator. Conventional energy discrimination is not possible with a continuous photon spectrum; therefore, the contribution of scatter is difficult to quantify in the image. In addition, since the incident photons range in energy well above 108 keV, there is considerable septal penetration and scatter within the collimator as well as the camera housing that further contribute to degraded image quality. With all of these effects, it is still possible to obtain good quality images of 90Y biodistribution post embolization with SPECT as shown in ▶ Fig. 5.18. Further advancements in SPECT quantification have worked toward verifying the fraction of administered activity in the liver but the physical complexity increases as corrections for both attenuation and scatter in soft tissue depend on photon energy.

Data were collected with 128 × 128 matrix and reconstructed with OSEM and no attenuation correction.

5.9.3 Discussion 90Y

emits a high-energy beta particle (2.28 MeV maximum) that undergoes bremsstrahlung

5.9.4 Resolution Imaging of bremsstrahlung photons can be performed with good image quality but careful consideration is needed when selecting an energy window and collimator as these factors affect image quality.

Fig. 5.17 Photon energy spectrums from unshielded 99mTc (a) and 90Y (b). The photopeak energy at 72–88 keV is a Pb characteristic X-ray from photoelectric interactions with the lead collimator.

110

| 09.11.19 - 01:53

5.9 Case 9: Bremsstrahlung Imaging of 90Y Microspheres Liver Emboliza

Fig. 5.18 99mTc microaggregated albumin (MAA) planning (middle panel) and 90Y bremsstrahlung (bottom panel) singlephoton emission computed tomography images fused with MR. Patient is a 71-year-old male. 4 mCi 99mTc MAA and 28.6 mCi 90 SIR-Sphere administrations.

111

| 09.11.19 - 01:53

Nuclear Medicine

5.10 Case 10: Degraded Image Quality from an Improper Collimator 5.10.1 Background ●

Patient screened for bone metastases with known cancer.

●

25 mCi of 99mTc MDP was administered.

●

Whole-body planar imaging was performed about 45 minute after MDP administration (▶ Fig. 5.19).

5.10.2 Findings ●

Images are of inferior quality compared to previous patient images (▶ Fig. 5.20).

●

The image resolution appears worse than typical, which is more evident in the posterior planar view.

5.10.3 Discussion Review of the protocol and patient setup revealed a technical error where the mediumenergy collimators were used during imaging. The technologist operating the camera had previously performed imaging with 111In (photo-

peaks: 171 keV, 245 keV) and medium-energy collimators. LEHR collimators are most commonly used with 99mTc-labeled radiopharmaceuticals and are specifically designed to optimize resolution and sensitivity of 140 keV photons from 99mTc. Compared to Low energy high resolution (LEHR), medium-energy collimators typically have larger holes and thicker septa in order to maintain sensitivity and prevent resolution degradation from septal penetration when imaging with photon energies above approximately 200 keV. The consequence of using medium-energy collimators with 99mTc is a loss of resolution due to the larger collimator hole aperture as illustrated in ▶ Fig. 5.21. The wider angle of acceptance allows more photons to reach the detector crystal from oblique angles. This leads to a degraded contrast and spatial resolution for objects located at the same distance compared to imaging with a LEHR collimator.

5.10.4 Resolution Loss of resolution from this technical error cannot be resolved with post-processing. The technologist should switch from medium-energy collimators to low-energy collimators and reimage the patient. A satisfactory bone scan with the proper collimators is shown in ▶ Fig. 5.10. Fig. 5.19 Repeat bone scan collected for evaluation of bone metastases (right). Prior bone scan on the same camera from a different patient (left). There is a degradation in image resolution from left to right that is also visible in the anterior view (not shown).

Fig. 5.20 Medium-energy collimators (left) and a repeat scan with lowenergy high-resolution collimators (right).

112

| 09.11.19 - 01:53

5.10 Case 10: Degraded Image Quality from an Improper Collimator

Fig. 5.21 Illustration of the effect of resolution for a medium- and lowenergy collimator for a source at the same distance from the collimator. The measure of point spread function of a point source at the same distance is wider for the medium-energy collimator compared to the low-energy collimator. Widening of the point spread function results in degraded resolution and contrast.

113

| 09.11.19 - 01:53

Nuclear Medicine

5.11 Review Questions 5.11.1 Case 1: Degraded Resolution of a Whole-Body Planar 99mTc Methylene Diphosphonate Image 1. A quality control source of Technetium-99 m (99mTc) yielding 1200 cps is placed 10 cm from the face of a gamma camera with a parallelhole collimator. If the source is move to 20 cm from the face of the gamma camera, what is the expected count rate? a) 1800 cps b) 1200 cps c) 800 cps d) 600 cps

c) Detector size d) Depth of interaction

5.11.3 Case 3: Standardized Uptake Value in Positron Emission Tomography (Resolution and Noise) 5. Calculate the standardized update value for patient weighing 80 kg who is administered 10 mCi of [18F]FDG with a decay corrected measured lesion concentration of 300 nCi/mL. a) 2.4 b) 3.2 c) 4.1 d) 5.3 6. The mean liver SUV from an [18F]FDG oncology

2. The parallel-hole collimator having 1 cm length

PET/CT study was measured to be 1.6. It was

holes is replaced with another parallel-hole

later discovered that the patient weight was

collimator of 1.5 cm length holes of the same

incorrectly documented during the exam as 180

septal thickness. This change will result in a

lb and should have been recorded as 230 lb.

reduction of what performance characteristic?

What is the mean liver SUV with this revised

a) Sensitivity

weight?

b) Resolution

a) 1.6

c) Magnification

b) 1.8

d) Energy resolution

c) 2.0 d) 2.2

5.11.2 Case 2: Effect of Positron Range on Image Quality and Resolution

5.11.4 Case 4: The Impact of Attenuation Correction in PET

3. The system response function, including all

7. Including attenuation correction, which other

resolution-degrading factors of detector size,

data correction process in PET relies on accu-

noncollinearity, and positron range, is meas-

rate measurement of the attenuation map?

ured by imaging what type of object?

a) Normalization

a) Point source

b) Scatter

b) Rotating rod source

c) Randoms

c) Large uniform phantom

d) Well-counter calibration

d) An anthropomorphic phantom 8. What is the primary interaction process of 4. Which resolution-degrading aspect of the PET

114

511 keV annihilation photons in the body?

image does not depend on the instrument’s

a) Coherent scattering

construction?

b) Compton scattering

a) Noncollinearity

c) Photoelectric absorption

b) Positron range

d) Pair production

| 09.11.19 - 01:53

5.11 Review Questions

5.11.5 Case 5: Iterative Reconstruction and Choosing the Number of Iterations and Subsets

5.11.7 Case 7: Choosing the Correct Acquisition Image Matrix Size

9. Which reconstruction approach is not based

13. What is the pixel size of a planar gamma

on an iterative algorithm?

acquisition with a 25-cm FOV collected with a

a) Maximum likelihood expectation

128 × 128 pixel matrix?

maximization

a) 1.20 mm

b) Conjugate gradient minimization

b) 1.95 mm

c) Algebraic reconstruction technique

c) 2.32 mm

d) Filtered backprojection

d) 3.21 mm

10. What is an advantage of using filtered backpro-

14. When switching from a 64 × 64 to a 128 × 128

jection over iterative reconstruction routines?

matrix, how many more counts are need

a) Increased speed

to maintain the same noise properties

b) Reduced noise

(× = times)?

c) Improved spatial resolution

a) 1 ×

d) Resolution modeling

b) 2 × c) 3 ×

5.11.6 Case 6: The Effects of Image Smoothing

d) 4 ×

reconstruction?

5.11.8 Case 8: Assessing Patient Motion in Myocardial Perfusion Imaging

a) Gaussian

15. When projection data are organized into a

11. Which image filtering type is commonly applied in the image space domain after

b) Butterworth

sinogram, patient motion along the table direc-

c) Shepp–Logan

tion leads to as what appearance in the sinus-

d) Ramp

oidal information? a) Area of high counts

12. Increasing the Butterworth cutoff frequency

b) Data discontinuities

toward the Nyquist limit changes which

c) Complete loss of data

characteristic of image quality?

d) No change can be observed

a) Reduces noise b) Improves low contrast

16. How does patient motion change image

c) Increases resolution

resolution?

d) Introduces aliasing

a) Degrades b) No change c) Improves

115

| 09.11.19 - 01:53

Nuclear Medicine

5.11.9 Case 9: Bremsstrahlung Imaging of 90Y Microspheres Liver Embolization 17. What photon interaction process produces the Pb characteristic X-rays commonly observed in a gamma camera energy spectrum? a) Photoelectric b) Compton c) Coherent d) Pair production 18. What is a common energy window width used in 99mTc radiopharmaceutical imaging? a) 10–15% b) 15–20% c) 20–25% d) 25–30%

5.11.10 Case 10 Degraded Image Quality from an Improper Collimator 19. What collimator is appropriate for performing a planar gallium-67 (Ga-67) citrate study? (Ga-67 has a photopeak at 93 keV (37%), 187 keV (20.4%), and 300 keV (16.6%)). a) Low-energy high-resolution b) Low-energy general-purpose c) Medium-energy d) High-energy 20. Increasing the collimator hole length has what effect on image quality? a) Reduction in noise b) Improved resolution c) Increased sensitivity d) Decreased scatter rejection

116

References [1] Naddaf SY, Collier BD, Elgazzar AH, Khalil MM. Technical errors in planar bone scanning. J Nucl Med Technol. 2004; 32 (3):148–153 [2] Cherry SR, Sorenson JA, Phelps ME. The gamma camera: basic principles. In: Cherry SR, Sorenson JA, Phelps ME, eds. Physics in Nuclear Medicine 4th ed. Philadelphia: W.B. Saunders; 2012:195–208 [3] Cherry SR, Sorenson JA, Phelps ME. The gamma camera: performance characteristics. In: Cherry SR, Sorenson JA, Phelps ME, eds. Physics in Nuclear Medicine. 4th ed. Philadelphia, PA: W.B. Saunders; 2012:209–31 [4] Cherry SR, Sorenson JA, Phelps ME. Positron emission tomography. In: Cherry SR, Sorenson JA, Phelps ME, eds. Physics in Nuclear Medicine. 4th ed. Philadelphia, PA: W.B. Saunders; 2012:307–43 [5] Votaw JR. The AAPM/RSNA physics tutorial for residents. Physics of PET. Radiographics. 1995; 15(5):1179–1190 [6] Kinahan PE, Fletcher JW. Positron emission tomographycomputed tomography standardized uptake values in clinical practice and assessing response to therapy. Semin Ultrasound CT MR. 2010; 31(6):496–505 [7] Thie JA. Understanding the standardized uptake value, its methods, and implications for usage. J Nucl Med. 2004; 45 (9):1431–1434 [8] Blodgett TM, Mehta AS, Mehta AS, Laymon CM, Carney J, Townsend DW. PET/CT artifacts. Clin Imaging. 2011; 35(1): 49–63 [9] Tong S, Alessio AM, Kinahan PE. Image reconstruction for PET/CT scanners: past achievements and future challenges. Imaging Med. 2010; 2(5):529–545 [10] Galt JR, Hise HL, Garcia EV, Nowak DJ. Filtering in frequency space. J Nucl Med Technol. 1986; 14(3):152–160 [11] Burrell S, MacDonald A. Artifacts and pitfalls in myocardial perfusion imaging. J Nucl Med Technol. 2006; 34(4):193– 211, quiz 212–214

| 09.11.19 - 01:53

6 Ultrasound Imaging Zheng Feng Lu

Introduction An ultrasound image is formed using sound. Ultrasound has been widely used because it is portable, low cost, and does not involve ionizing radiation; thus, it is safe even for scanning a fetus. B-mode ultrasound image formation is based upon three basic assumptions:1

Common Image Quality Problems ●

resolution and the accuracy of the distance measurements. ●

1. The sound travels in a straight and narrow line called an acoustic beam. A transducer is

Frequency: it affects spatial resolution and the maximum depth of penetration.

●

used both as a pulse emitter and as an echo

Array transducer dropouts: the deficiency most commonly found during routine quality control

receiver. The transducer emits a short pulse and then receives echo signals which are

Speed of sound propagation: it affects spatial

(QC) testing. ●

Presets in image protocols: optimize the image

generated by the emitted pulse propagating

acquisition controls and maintain the consis-

and interacting with tissue within the

tency by managing the “presets” in image

acoustic beam.

protocols.

2. The speed of the sound propagation is assumed

●

Acoustic window: it affects the coupling of the

to be constant at 1,540 m/s, an average speed

transducer and patient body. A poor acoustic

for soft tissues. Therefore, on the image, the

window prevents the sound from being trans-

sources for echo signals are localized by the

mitted to the region of interest.

so-called range equation:

●

ct 2

●

D¼

Where t is the time delay between the pulse emission and the echo reception, c represents the speed of sound (i.e., 1,540 m/s is assumed for soft tissue), and D is the depth of the echo where it is generated. The time delay t includes both the time it takes for the pulse to travel down to the reflector and the time it takes for the echo to return. Therefore, it accounts for the factor of 2.

ring-down can provide diagnostic information. Range ambiguity occurs when the pulse repetition frequency (PRF) is too high with the result that echoes from the prior beam line are mispositioned in the current beam line. ●

Enhancement and shadowing can provide diagnostic information.

●

Harmonic imaging is advantageous over conventional B-mode imaging in generating superior image quality, especially in the case of a “techni-

3. The received echo signals are amplified to compensate for the attenuation. The strength

Reverberation artifacts such as comet-tail and

cally difficult” patient with a thick body wall. ●

The performance evaluation of the ultrasound

of the echo signals is represented by varying

scanner display and the reading room worksta-

brightness levels on a B-mode image.

tion displays is crucial in maintaining consis-

Ultrasound images are formed using the idealized model described above. Ultrasound artifacts occur when any or all of the assumptions are violated.

tency of image perception. ●

Doppler aliasing occurs when the PRF is too low, a control that is linked to the PRF. To reduce aliasing, one should increase the velocity scale.

117

| 09.11.19 - 01:53

Ultrasound Imaging

6.1 Case 1: Pulse-Echo Imaging Principle and Speed of Sound Propagation

The misestimated speed of sound propagation affects the image quality as demonstrated by the group of pins in ▶ Fig. 6.2.

6.1.1 Background

6.1.3 Discussion

Size measurement is a routine clinical application in ultrasound imaging. The range equation is implicitly built into pulse-echo imaging instruments to localize the received echo signals. If the sound propagation speed in the body tissue is indeed 1540 m/s, the measured size will be accurate. However, if the sound propagation speed is less than 1540 m/s, then the longer delay in echo return time is interpreted by longer distance; in this case the distance in the axial direction is overestimated. On the other hand, if the speed is greater than 1540 m/s, then the distance in the axial direction is underestimated. The misestimated sound propagation speed not only affects the distance measurement, but also contributes to the reduction of the spatial resolution; thus, it worsens the image quality.

The speed of sound propagation is dependent upon the tissue properties such as density and compressibility. For any particular material, the higher its density and the harder it is to compress, the faster its sound speed will be. For example, the sound speed is faster in bone than that in soft tissue because bone is more dense and harder to compress than soft tissue. ▶ Table 6.1 shows the typical speed of sound propagation in a variety of materials. In diagnostic ultrasound, a sound speed of 1,540 m/s is assumed. This value represents the average speed of sound propagation in soft tissues. Any deviation from the assumed speed causes errors in size measurements, known as speed artifacts. As the acoustic beam focusing relies on the assumption of the sound speed, image quality is affected by any deviation in the actual sound speed and the assumed sound speed by the ultrasound system in setting up the beam former. As demonstrated in ▶ Fig. 6.2, the beam focuses the best when the assumed sound speed matches the actual sound speed at 1460 m/s.

6.1.2 Findings In the following clinical case, the discontinuity seen in the diaphragm pointed by the arrow in ▶ Fig. 6.1a is caused by the different sound speed in the lesion. The sound speed reduces in a fatty lesion. Therefore, the portion of the diaphragm (pointed by the arrow) below the lesion is shown in a deeper position in the image. In another clinical case shown in ▶ Fig. 6.1b, an irregular liver/diaphragm interface is seen in the image caused by heterogeneous liver parenchyma with fatty infiltration.

6.1.4 Resolution It is important to recognize potential speed artifacts by examining the acoustic pathway and identifying regions that are suspected to have different values for sound speed. Typically,

Fig. 6.1 Speed artifacts shown as the discontinuity at the liver/diaphragm interface (pointed by the arrow) caused by a lesion (a). An irregular liver/diaphragm interface shown in the image caused by heterogeneous liver parenchyma with fatty infiltration (b).

118

| 09.11.19 - 01:53

6.1 Case 1: Pulse-Echo Imaging Principle and Speed of Sound Propagation

Fig. 6.2 (a–d) A set of images of a group of pins in a urethane phantom under different settings of speed of sound propagation. The actual speed of sound in urethane is 1,450 m/s as per manufacturer’s specification. The spatial resolution is at its best when the speed is set at 1,460 m/s that is the closest to the actual speed in urethane material.

Table 6.1 Sound propagation speed in selected materials with 1,540 m/s as the average speed for soft tissue2 Media

Lung

Fat

Water

Liver

Blood

Kidney

Muscle

Skull bone

Speed (m/s)

600

1,460

1,498 at 25 °C

1,555

1,560

1,565

1,600

4,080

nothing is done to correct for speed artifacts. However, some modern ultrasound systems allowmanual adjustment to correct for deviations in the speed of sound in order to improve image quality. As demonstrated in ▶ Fig. 6.2, the actual speed of sound of the urethane-based phantom is approximately 6% lower than 1,540 m/s.3 When the machine-assumed speed of sound does not match the actual speed of sound in the urethane-based phantom, the lateral resolution deteriorates. This feature has been used for breast imaging. As shown in ▶ Fig. 6.3, the image quality is improved by an adjustment from the typical sound speed, i.e., 1,540 m/s to the actual sound speed in breast tissue at 1,500 m/s.

Fig. 6.3 A breast image with the speed of sound propagation set at 1,500 m/s to match the actual speed of sound in breast tissue to improve image quality.

119

| 09.11.19 - 01:53

Ultrasound Imaging

6.2 Case 2: Array Transducers and Sound Frequency

Linear Array ●

acoustic beam perpendicular to the face of the

6.2.1 Background Sound frequency is the number of oscillations per unit time, determined by the source—ultrasound transducer. For human ears, the audible sound frequency ranges from 20 Hz to 20 kHz. Sound with frequencies above 20 kHz is called ultrasound. The higher the frequency, the better the spatial resolution in ultrasound images. On the other hand, the lower the frequency, the deeper the ultrasound penetrates because the ultrasound attenuation is proportional to frequency. In medical ultrasound imaging, the frequency typically ranges from 2 to 15 MHz with low frequency used for imaging deep targets but with compromised spatial resolution, and high frequency used for imaging superficial targets with better spatial resolution. High frequencies, around 50 MHz or even higher, are used for certain specialized imaging applications, such as in ophthalmology or in small animal imaging, with superb spatial resolution. An ultrasound system typically has multiple transducers of various frequencies, each transducer having a different contact surface size (footprint) designed for specific clinical applications. Modern ultrasound systems often allow the operator to choose the frequency within a range on the same transducer, which allows for adjustment between penetration and spatial resolution.

6.2.2 Findings Different types of array transducers and their corresponding image examples are shown in ▶ Fig. 6.4.

A subgroup of elements is fired to form one array.

●

Subgroups of elements are fired sequentially to form parallel acoustic beams to scan across a region of interest (ROI).

●

Rectangular shape of the image format.

●

Clinical applications include small parts, vascular, and obstetric exams (▶ Fig. 6.4a).

Curvilinear or Curved Array ●

Same as linear arrays except that the elements are aligned in a curve instead of a straight line.

●

Each acoustic beam is perpendicular to the face of the array. Since the array of elements is curved, the acoustic beams diverge with depth which allows for a broader coverage as the depth increases.

●

Clinical applications include general abdominal, obstetric, and transabdominal pelvic exams. In addition, transvaginal and transrectal probes are curved arrays (▶ Fig. 6.4b).

Phased Array ●

All the elements are fired to form one acoustic beam.

●

The acoustic beam is electronically steered across the ROI.

●

The footprint is typically very small to allow for a small acoustic window (e.g., between the ribs)

Fig. 6.4 Various types of array transducers and their corresponding image examples. (a) An image example of a linear array transducer. (b) An image example of a curvilinear array transducer. (c) An image example of a phased array transducer.

120

| 09.11.19 - 01:53

6.2 Case 2: Array Transducers and Sound Frequency but the sector image format allows for a broad field of view. ●

Clinical applications include intercostal scanning for heart, liver, or spleen (▶ Fig. 6.4c).

6.2.3 Discussion On modern ultrasound systems, frequency can be selected within a certain range on the same transducer. While making the frequency selection, one must consider the fundamental trade-off between spatial resolution and penetration, and strike an optimal balance between these two factors (▶ Fig. 6.5).

6.2.4 Resolution It is possible to improve penetration at high frequency with advanced technology such as coded excitation technology. This technology plants a code in the transmitted pulses. In return, the received echo signals are carrying the same code; thus, raising the ability to differentiate between echo signals and noises. Through appropriate coding on transmitted pulses and decoding on received echo signals, one can improve the signalto-noise ratio and can, therefore, mitigate the trade-off between better spatial resolution, which is associated with higher frequency, and less penetration, also associated with higher frequency.

Fig. 6.5 A phantom is scanned by the same transducer with different frequency selections.

121

| 09.11.19 - 01:53

Ultrasound Imaging

6.3 Case 3: Nonuniformity (Array Transducer Element Dropouts) 6.3.1 Background An ultrasound array transducer contains an array (or arrays) of composite ceramic piezoelectric elements connected by wires enclosed in a cable that runs to the ultrasound system via a connector. An ultrasound transducer is vulnerable to damage because it is handled frequently during ultrasound imaging operation and may easily be dropped or bumped against hard surfaces, or its cable may be rolled underneath the scanner wheels during transportation for portable studies. An ultrasound transducer is prone to image nonuniformity problems due to any of the following defects: (1) failed crystal elements; (2) delamination of the lens/coupling layers on the transducer face; (3) broken wires in the transducer cable; and (4) disruptions in the connector. A picture of transducers with tangled cables connected to an ultrasound system is shown in ▶ Fig. 6.6a. Transducers are often hung on walls, as shown in ▶ Fig. 6.6b, c. Too much stress on a transducer cable may tear

the wires inside the cable and cause element dropouts. Handling transducers with care is critical to ensuring the longevity of these array transducers.

6.3.2 Findings Sometimes the nonuniformities are minor, appearing as streaks along the axial direction of the transducer, whereas some are more prominent. ▶ Fig. 6.7 shows the images generated by the same transducer. Nonuniformities can be seen on the clinical image and the phantom image, as well as on the in-air scan image.

6.3.3 Discussion Image nonuniformity is considered the most commonly found deficiency during routine QC testing.4 As shown in the example in ▶ Fig. 6.8, the transducer cable was accidentally rolled underneath the scanner wheel and five wires in the cable were broken causing minor dark streaks near the face of the transducer. The broken wires were detected by an electronic transducer testing device.5 The nonuniformity was much less perceivable on clinical images.

Fig. 6.6 Ultrasound transducers are handled frequently during ultrasound imaging operation. Education on careful handling of the transducers can reduce transducer failure rate. (a) Transducers should be hung properly on the system to avoid tangles or being trapped under the wheels of the system. (b) While it is neat to hang up transducers, too much stress on transducer cable may tear the wires inside the cable and cause element dropouts. (c) Various transducer hanging boxes such as this one are designed to handle transducers with care by minimizing the stress on transducer probe, its connector and cable.

122

| 09.11.19 - 02:58

6.3 Case 3: Nonuniformity (Array Transducer Element Dropouts)

Fig. 6.7 (a–c) Nonuniformities are seen in the images of this transducer (pointed by arrows) on a patient, a phantom, and just in-air.

Fig. 6.8 Both images were obtained by the same transducer. (a) The streaks observed near the face of the transducer in the phantom image are hard to see in the (b) clinical image.

6.3.4 Resolution It is important to perform periodic QC testing using an ultrasound QC phantom to detect the presence of any nonuniformities as early as possible in order to monitor the condition and address the problem

before clinical applications are impacted. The image uniformity test is a required QC test by American College of Radiology (ACR) Ultrasound Accreditation Program for all transducers used for clinical applications.6

123

| 09.11.19 - 02:58

Ultrasound Imaging

6.4 Case 4: Pulse-Echo Imaging Acquisition Controls 6.4.1 Background

6.4.3 Discussion Four ultrasound pulse-echo imaging acquisition parameters are explained below. ●

The pulse-echo imaging controls on an ultrasound system affect both image data acquisition and image display. The proper setting of these controls is crucial for optimizing the image performance to fulfill the clinical task.

Transmit power: This determines the amplitude of the pulse emitted from the transducer. Higher transmit power will yield stronger returned echoes, thus deeper penetration depth as demonstrated in ▶ Fig. 6.9. This control affects the acoustic dosimetry that is typically indicated by the

6.4.2 Findings We will discuss the following four main controls:

mechanical index (MI) and the thermal index (TI). ●

Gain (overall gain): This amplifies the echo signals of all depths. It is often used to adjust the overall

●

Transmit power (▶ Fig. 6.9)

image brightness. Unlike transmit power control,

●

Overall gain (▶ Fig. 6.10 and ▶ Fig. 6.11)

this control does not affect the penetration depth,

●

Time gain compensation (TGC) (▶ Fig. 6.12)

as demonstrated in ▶ Fig. 6.10 and ▶ Fig. 6.11.

●

Dynamic range (▶ Fig. 6.13)

Neither does it affect the acoustic dosimetry.

Fig. 6.9 The ultrasound phantom images were acquired under identical instrumentation settings except the transmit power setting at maximum in (a), − 6 dB in (b) and − 12 dB in (c). As the transmit power reduces, the image becomes dimmer and with less penetration.

Fig. 6.10 The ultrasound phantom images were acquired under identical instrumentation settings except the overall gain setting at 100% in (a), 70% in (b), and 60% in (c). As the overall gain reduces, the image becomes dimmer but the penetration remains the same.

124

| 09.11.19 - 01:53

6.4 Case 4: Pulse-Echo Imaging Acquisition Controls

Fig. 6.11 The ultrasound phantom images were acquired under identical instrumentation settings except that the overall gain setting is increased by 5 dB at each increment from 30 dB at the upper left panel to 70 dB at the lower right panel.

Fig. 6.12 The two ultrasound phantom images were acquired under identical instrumentation settings except that the time gain compensation (TGC) setting is set at the center position in the upper panel and one of the TGC knobs was set at its minimum position in the lower panel.

125

| 09.11.19 - 01:53

Ultrasound Imaging

Fig. 6.13 (a) The ultrasound images were acquired under identical instrumentation settings except the dynamic range setting at 36 dB, (b) 60 dB, and (c) 96 dB. As the dynamic range increases, the image becomes smoother but with less contrast.

●

TGC: Due to ultrasound attenuation, echoes

dynamic range indicates the range from the

returning to the transducer get weaker from the

smallest to the largest echo signals to which the

distance traveled. The longer distance it travels,

system can properly respond. It is described in

the weaker the echo signal becomes. Amplifica-

decibels. A smaller dynamic range setting

tion in the receiver, called TGC, is needed to offset

means a steeper gradient that provides more

the loss due to attenuation as the depth increases.

contrast on the image display, increasing the

By increasing amplification along with the depth,

conspicuity of a low-contrast lesion. However,

TGC offers a more uniform display of the bright-

the image will appear coarse. Conversely, a

ness level throughout the field of view (FOV). TGC

larger dynamic range setting has a lower

can be altered by a group of sliding knobs, each of

contrast gradient, but the image will appear

which amplifies the echo signals from a specific

smoother. This is demonstrated in ▶ Fig. 6.13.

depth range. The TGC knobs are calibrated in such a way that when all the knobs are aligned in the center position, the image of an organ, for example, a liver, appears with uniform brightness at all depths as demonstrated in ▶ Fig. 6.12. ●

Dynamic range: In ultrasound imaging, echo signals are logarithmically compressed and transformed by decibel notation, defined as 10 times the log10 of the ratio of echo signal intensity compared to reference intensity. The

126

6.4.4 Resolution The ultrasound image acquisition controls, discussed here, are user adjustable. Typically, optimization of these controls is conducted through “presets” on the system protocol and may vary per transducer and per body part, even for an ultrasound system of the same vendor and same model. Therefore, understanding and managing these controls helps maintain the consistency of the imaging performance.

| 09.11.19 - 01:53

6.5 Case 5: Reflection (Boundary Conditions)—Reverberation Artifacts

6.5 Case 5: Reflection (Boundary Conditions)— Reverberation Artifacts 6.5.1 Background A sound beam is reflected when it is incident on an interface formed by two tissues of different acoustic impedance. The magnitude of the reflection depends on the acoustic impedance difference at the interface. The smaller the acoustic impedance difference, the less the reflected beam energy, and the better the transmission of the sound beam will be. Acoustic impedance is a property of the tissue that equals to the product of tissue density and the sound propagation speed. Therefore, the acoustic impedance of soft tissue is very different from the acoustic impedance of bone or gas. A good acoustic window is a body location that allows minimal sound reflection and optimal sound transmission, for example, region with no bone or gas to compromise sound transmission.

6.5.2 Findings Ultrasound imaging assumes that an echo returns to the transducer after a single reflection. Reverberation artifacts occur when this assumption fails, as shown in ▶ Fig. 6.14.

6.5.3 Discussion When two highly reflective interfaces run parallel in the beam propagation path, it may

Fig. 6.14 Reverberation artifacts occur due to parallel reflective interfaces in the body wall.

result in multiple reflections back and forth between the two interfaces. The first received echo will appear at the proper depth on the image, but the subsequent echo signals received from the multiple reflections between the two interfaces will appear at deeper depths; the extra time delays caused by multiple reflections are interpreted by the system as representing longer distances traveled by the sound. These additional echo signals appear equally spaced on the image but with decreasing intensity, as each reflection is weaker than the prior. Comet-tail artifact as shown in ▶ Fig. 6.15 is a special subtype of reverberation artifact which occurs when the two highly reflective interfaces are too close to each other to be resolvable on the image. As implied by its name, comet-tail artifact looks like a comet tail with a tapering echogenic triangle-shaped tail. It often appears in response to a highly reflective object such as cholesterol crystals in adenomyomatosis of the gallbladder or a bullet fragment. Ring-down artifact as shown in ▶ Fig. 6.16 is often included as another special subtype of reverberation artifacts. Although it may resemble a comet-tail artifact, the ring-down artifact is different. Caused by resonant vibrations within the center of a cluster of air bubbles, a ring-down artifact appears as a long solid streak or series of steaks along the sound propagation axis. In a 1985 publication,7 Avruch et al demonstrated that ring-down never occurred with only one layer of air bubbles. Rather, it only resonated from a fluid center trapped within a cluster or tetrahedron of air bubbles: three air bubbles on top and one air bubble nestled beneath.

Fig. 6.15 Breast imaging with surgical clips showing comet-tail artifacts.

127

| 09.11.19 - 01:53

Ultrasound Imaging

6.5.4 Resolution

Fig. 6.16 Ring-down artifact caused by air bubbles in the bowel.

128

The reverberation artifacts are easy to identify, as shown in ▶ Fig. 6.14, ▶ Fig. 6.15, and ▶ Fig. 6.16. Typically, nothing is done about it. Sometimes, a different acoustic window is chosen to avoid severe reverberation artifacts. Comet-tail artifacts are clinically useful because small objects, such as surgical clips, can be identified through the occurrence of comet-tail artifacts, as shown in ▶ Fig. 6.15. Ringdown artifact can be useful in providing diagnostic information, for example, in case of emphysematous (gas-forming) infections and abscesses that often produce ring-down artifacts.

| 09.11.19 - 01:53

6.6 Case 6: Range Ambiguity in B-Mode

6.6 Case 6: Range Ambiguity in B-Mode

the above mentioned assumption fails. This is called the range ambiguity artifact.

6.6.1 Background

6.6.3 Discussion

The conventional B-mode image is formed by ultrasound beam line, one after another. For each ultrasound beam line, a very short ultrasound pulse (< 1 μs) is emitted from the transducer and then echo signals are received along the beam line. The next pulse will not be emitted until after the echo signal from the deepest range of the FOV from the prior line is received. The time between successive individual pulse emissions is called the pulse repetition period (PRP). As a result, the deeper the range of the FOV, the longer the PRP is needed.

If an echo from a distant structure is received after the next pulse is transmitted, i.e., beyond the PRP, the time delay will be counted from the second pulse emission instead of the first pulse emission. Consequently, the distance will be mispositioned to be closer to the transducer than it actually is. Any scanner parameter setting that shortens the time interval between pulse emissions, for example, setting up multiple focal zones, is susceptible to range ambiguity artifacts.

6.6.2 Findings Pulse-echo image formation assumes that returning echo signals are all generated by the latest pulse emission. As shown in ▶ Fig. 6.17, a horizontal line formed by reflection echoes from the bottom of the phantom is mispositioned when

6.6.4 Resolution The name “range ambiguity” refers to uncertainties in the actual range from where the echo signal occurs. When it occurs, it is likely visible in large fluid-filled structure, misleading to mimic debris in the structure. Range ambiguity artifact can be minimized by allowing more time for echoes from deeper structures to arrive before firing the next pulse.

Fig. 6.17 Range ambiguity artifact is shown in this phantom image (pointed by the arrow). The horizontal line formed by reflection echoes from the bottom of the phantom is mispositioned. This occurred when the reflection echo from the bottom of the phantom was received after the next pulse was transmitted; and was thus counted as echo for the next beam line.

129

| 09.11.19 - 01:53

Ultrasound Imaging

6.7 Case 7: Shadowing and Enhancement (Increased Through Transmission) 6.7.1 Background As an ultrasound wave propagates through tissue, its intensity is reduced over distance due to absorption, scattering, and reflection. The ultrasound attenuation coefficient is given in decibels per centimeter (dB/cm).

6.7.2 Findings Ultrasound attenuation coefficient is an important characteristic property of a material. For example, a fatty infiltrated liver tissue has a higher attenuation coefficient than a healthy liver tissue. For a focal lesion with higher attenuation coefficient in comparison to that of its surrounding tissue, shadowing occurs (▶ Fig. 6.18). Otherwise, enhancement occurs (▶ Fig. 6.19).

6.7.3 Discussion Reduced echo intensity appears as “shadowing” distal to a highly attenuating or reflective object such as a tumor (▶ Fig. 6.18). On the other hand, increased echo intensity appears as “enhancement” distal to an object with lower attenuation such as a fluid-filled gallbladder (▶ Fig. 6.19).

6.7.4 Resolution Shadowing and enhancement are considered useful image artifacts because these artifacts indicate the attenuation properties of the object that causes the occurrence of shadowing or enhancement. For example, enhancement is often used to differentiate cystic structures from solid structures and shadowing is used to detect stones, calcified objects, and air. One must be careful that spatial compounding, an advanced imaging feature, may affect the appearance of shadowing or enhancement. When spatial compounding is activated, each ultrasound beam is steered into different angles and multiple

Fig. 6.18 Shadowing is shown behind the breast mass due to higher attenuation in the mass in comparison to the surrounding tissue.

Fig. 6.19 Enhancement (increased through transmission) is shown behind the gallbladder due to lower attenuation in the gallbladder in comparison to the surrounding tissue.

130

| 09.11.19 - 01:53

6.7 Case 7: Shadowing and Enhancement (Increased Through Transmission) steered frames are rendered to create an image with less speckle and better signal-to-noise ratio. However, as the spatial compounding steers the ultrasound beam, the shadowing or enhancement diverges and loses its intensity, thus becoming less noticeable. Since spatial compounding is typically activated in the preset

image protocol, its effect on shadowing and enhancement needs to be understood, especially when the object in question is small. For example, to see the shadowing associated with a kidney stone, the operator should deactivate the spatial compounding feature.

131

| 09.11.19 - 01:53

Ultrasound Imaging

Harmonic imaging is superior in image quality. A comparison of harmonic image and conventional B-mode image is shown in ▶ Fig. 6.20.

fundamental, or in this case, 7 MHz. When a transducer transmits a burst of ultrasound at a fundamental frequency, the sound wave gradually distorts as it propagates due to the fact that the compressional part of the wave travels slightly faster than the rarefactional part. Consequently, this distortion is accompanied by the generation of harmonics that can be used to form images. This process is called harmonic imaging. The benefits of using the harmonic imaging include improved contrast resolution, reduced clutter, improved spatial resolution, and reduced section thickness.8 As shown in ▶ Fig. 6.20, THI has superior border and tissue definition with reduced speckles.

6.8.3 Discussion

6.8.4 Resolution

The term “harmonic” refers to those frequencies that are integral multiples of the transmitted frequency. The transmitted frequency is called the fundamental frequency or the first harmonic frequency. For example, for a 3.5-MHz transmitted pulse, its fundamental frequency, or the first harmonic frequency, is 3.5 MHz. Its second harmonic frequency is twice the

Many ultrasound imaging protocols have “harmonic imaging” in the preset as the default imaging mode instead of conventional B-mode. This is particularly helpful in imaging “technically challenging” patients who have thick body walls or other complicated structures that give rise to artifacts and clutters.

6.8 Case 8: Harmonic Imaging 6.8.1 Background Harmonic imaging was originally developed on the basis of nonlinear properties of sound propagation in ultrasound contrast agents. Later, it was revealed that the nonlinear effect was also present in tissues. Tissue harmonic imaging (THI) was developed based upon the nonlinear effect present in tissue.8

6.8.2 Findings

Fig. 6.20 (a) A comparison of a liver image acquired in B-mode (left) and the same patient scanned in tissue harmonic imaging mode (THI; right) is shown. Superior image quality is demonstrated with THI. (b) This panel shows a comparison of a phantom image acquired in B-mode (left) and the same phantom scanned in THI (right). Mirror artifact (pointed by arrows) is more prominent in the B-mode image.

132

| 09.11.19 - 01:53

6.9 Case 9: Ultrasound Image Display on Scanners and in Reading Rooms

6.9 Case 9: Ultrasound Image Display on Scanners and in Reading Rooms 6.9.1 Background For diagnostic ultrasound imaging, the display characteristics must be optimal in order to convey all details and features of ultrasound images to the human interpretter. In practice, if the operator of the ultrasound scanner cannot visualize a pathology on the scanner display, then the image cannot be properly acquired, and thus will not be sent to the picture archiving and communication system (PACS) and interpreted on reading room displays. Therefore, an ultrasound scanner display belongs to the category of diagnostic displays, just like the reading room displays. Ultrasound scanner display performance testing is required by the ACR Ultrasound Accreditation Program.6 In addition, consistency in image presentation between the

scanner monitor display and the reading room workstation monitor display must be verified.

6.9.2 Findings Matching the presentation on the ultrasound scanner display and the reading room workstation displays can be challenging as shown in ▶ Fig. 6.21.

6.9.3 Discussion When a brand new ultrasound scanner is installed, image presentation consistency between the scanner monitor display and the reading room workstation monitor display must be verified. The ultrasound system, of which the images are shown in ▶ Fig. 6.21, may have many different curves to export ultrasound images from the scanner to PACS. The pixel values of the image can be altered in order to match the ultrasound image presentation on the PACS display.

Fig. 6.21 For an ultrasound scanner, the setting of the system configuration can export the same ultrasound image to reading room in multiple different ways, resulting different appearances on reading room workstation display as shown here. LUT, look-up-table.

133

| 09.11.19 - 01:53

Ultrasound Imaging The limitation of this approach is that it is tailored to a specific characteristic display during the matching process but not generalizable to other displays with different characteristic curves. What makes establishing presentation consistency between the scanner display and the PACS display even more challenging is the fact that some ultrasound manufacturers have created special display look-up-tables (LUT) on their scanner display devices to enhance the ultrasound images. Such image enhancement cannot be replicated downstream on the PACS display because it is only available on the scanner display.

134

6.9.4 Resolution Standardized display systems are needed to map the presentation values to monitor luminance regardless of the display monitor or vendor software. Such a standardized display system, called the grayscale standard display function (GSDF),9 has already been developed by the digital imaging and communication (DICOM). As long as both the ultrasound scanner monitor and the reading room monitor follow GSDF and all the image enhancement is done in the presentation values, there is a good chance of preserving presentation consistency.

| 09.11.19 - 01:53

6.10 Case 10: Doppler Ultrasound Aliasing

6.10 Case 10: Doppler Ultrasound Aliasing 6.10.1 Background The capability to quickly image and quantify blood flow is unique to diagnostic ultrasound imaging. The physics principle is straightforward. Whenever there is a relative motion between a sound source and a listener, the frequency received by the listener is shifted from the frequency emitted by the source. The perceived frequency shift is called the Doppler shift. We experience the Doppler shift in routine life. For example, the horn of a train is shifted to a higher pitch when the train is approaching the platform and shifted to a lower pitch when the train is leaving the platform. Mathematically, the Doppler shift fD, the difference between the received sound frequency fr and the emitted sound frequency fo, is expressed as following for a blood flow (▶ Fig. 6.22) with Doppler angle : fD ¼fr  fo ≅

2vcosðÞf 0 c

The Doppler angle  is the angle between the sound beam and the flow direction. The Doppler angle affects the detected Doppler shift. When the Doppler angle is at 90 degrees, no Doppler shift can be detected. The rule of thumb in clinical applications is to keep the Doppler angle below 60 degrees.

6.10.2 Findings Doppler aliasing in spectral Doppler is shown in ▶ Fig. 6.23. The aliasing is manifested as "wraps around" from the bottom to the top in spectrum. Similarly, aliasing can also occur in color Doppler, showing the reversed color at the peak flow speed as if the flow direction is reversed (see ▶ Fig. 6.24).

6.10.3 Discussion The Doppler shift is detected based upon “sampled” echo signals. Each time a pulse is emitted and echo signals are collected along the beam line, the echo signals are sampled for Doppler shift analysis. Therefore, the sampling rate of a pulsed Doppler instrumentation is equal to the pulse repetition frequency (PRF). The greater the PRF, the better the rendition of the Doppler shift signals will be. If the PRF is less than twice the frequency of the maximum Doppler signal frequency, then aliasing will occur. The condition of two times the maximum Doppler shift is known as the Nyquist criteria. On a pulsed Doppler instrument, the PRF needs to be more than or equal to the Nyquist criteria frequency to prevent aliasing from happening.

6.10.4 Resolution To eliminate aliasing, the operator can take the following steps: Fig. 6.22 Illustration of detecting the Doppler shift from a flow with a flow velocity of V, sound speed c, emitting frequency f0 and receiving frequency fr.

135

| 09.11.19 - 01:53

Ultrasound Imaging

Fig. 6.23 Aliasing on a spectral Doppler display and aliasing elimination steps.

136

| 09.11.19 - 01:53

6.10 Case 10: Doppler Ultrasound Aliasing

Fig. 6.24 Aliasing on a color Doppler display. The color at the peak speed of the flow is changed from blue to yellow as if the flow direction is reversed (arrow).

●

●

●

Adjust the scale. The PRF is linked to the scale

●

If none of the above steps are successful, increase

setting. As the scale increases, the PRF increases.

the Doppler angle. Increasing the Doppler angle

Adjust the baseline if the scale has reached its

will lower the Doppler shift, thus lowering the

maximum and the flow is mainly in a single

Nyquist criteria. However, raising the Doppler

direction.

angle increases the uncertainty in the accuracy

If neither of the above two steps are effective in

of Doppler shift measurement and is thus not

eliminating aliasing, lower the transducer

desirable.

frequency. Lowering the transducer frequency will lower the Doppler shift, thus lowering the

Having a proper Doppler scale is crucial for good practices using Doppler ultrasound.

Nyquist criteria as well.

137

| 09.11.19 - 01:53

Ultrasound Imaging

6.11 Review Questions 6.11.1 Case 1: Pulse-Echo Imaging Principle and Speed of Sound Propagation 1. If the liver speed of sound is slower than 1,540 m/s, the size measured by the calipers in the following image would:

a) Linear array b) Curved array c) Phased array d) Mechanical sector 4. With identical instrumentation settings except the frequency selection by the same transducer, which image was generated at the highest frequency? a) Panel A

a) Be overestimated b) Be underestimated c) Be the same b) Panel B 2. The speed of the sound propagation is largely determined by which of the following factor? a) The transducer frequency b) The transmission power c) The medium stiffness d) The medium attenuation

6.11.2 Case 2: Array Transducers and Sound Frequency 3. What transducer was used for the following image?

138

c) Panel C

| 09.11.19 - 01:53

6.11 Review Questions d) Panel D

6. What QC test should be done to investigate the cause of the dark streak in the following image if the streak at the same transducer face location appears on many clinical images?

6.11.3 Case 3: Nonuniformity (Array Transducer Element Dropouts) 5. The hypoechoic streak indicated by the arrow

a) Maximum depth of penetration b) Distance accuracy

in the following image is most likely caused

c) Image uniformity

by?

d) Low contrast detectability

6.11.4 Case 4: Pulse-Echo Imaging Acquisition Controls 7. What acquisition control should be increased if more depth of penetration is needed? a) Transducer frequency b) Dynamic range c) Overall gain d) Transmit power

8. What acquisition control should be adjusted if a horizontal band of darkness is seen? a) Improper gain setting

a) TGC

b) Compound imaging

b) Dynamic range

c) Dead transducer elements

c) Overall gain

d) Inappropriate transducer pressure

d) Transmit power

139

| 09.11.19 - 01:53

Ultrasound Imaging

6.11.5 Case 5: Reflection (Boundary Conditions)— Reverberation Artifacts 9. What is the name of the artifact indicated by the following image?

c) Reduce transmit power d) Reduce transducer frequency

6.11.7 Case 7: Shadowing and Enhancement (Increased Through Transmission) 13. What is the name of the artifact indicated by the arrow in the following image?

a) Enhancement b) Comet tail c) Beam width d) Range ambiguity 10. What is the cause for the artifact indicated by the arrow the image from the previous question?

a) Shadowing

a) Multiple refractions

b) Comet-tail

b) Multiple reflections

c) Beam width

c) Beam steering

d) Ring-down

d) Beam diverging 14. How is the attenuation of the thyroid nodule

6.11.6 Case 6: Range Ambiguity in B-Mode

compared with the attenuation of the

11. Which of the following is most likely to occur when pulse repetition period is too short and an echo signal from the prior beam line is mispositioned in the current beam line? a) Speed artifact b) Aliasing c) Mirror artifact d) Range ambiguity 12. What can be done to minimize range ambiguity artifacts? a) Reduce overall gain b) Reduce number of focal zones

140

surrounding tissue in the following image? a) The nodule is more attenuating b) The nodule is less attenuating c) the same

| 09.11.19 - 01:53

6.11 Review Questions

6.11.8 Case 8: Harmonic Imaging 15. Which of the following is a benefit using the THI instead of conventional B-mode imaging? a) Improved depth of penetration

b) Scanner display performance c) Primary interpretation display performance d) Contrast resolution

c) Increased temporal resolution

6.11.10 Case 10: Doppler Ultrasound Aliasing

d) Reduced sound frequency

19. What is of the artifact indicated by the arrow in

b) Reduced clutter

the following image? 16. Which ultrasound imaging technology was effective to improve image quality and reduce artifact in abdominal ultrasound imaging of a patient with a very thick body wall? a) Spatial compounding b) Ultrafast imaging c) Harmonic imaging d) Coded excitation

6.11.9 Case 9: Ultrasound Image Display on Scanners and in Reading Rooms 17. Advantages of enabling the GSDF on ultrasound

a) Aliasing b) Spectral broadening

scanner display include all of the following

c) Twinkle

except:

d) Flash

a) Image enhancement performed by the ultrasound scanner display b) No custom matching needed on the reading room workstation displays c) Presentation consistency among various displays in the imaging chain

20. What can be done to minimize aliasing? a) Reduce wall filter b) Reduce gain c) Increase scale d) Increase frequency

d) More streamlined performance evaluation and quality control 18. Which of the following is a required QC test by the ACR Ultrasound Accreditation Program?

References [1] Zagzebski JA. Pulse-echo ultrasound instrumentation. Essentials of Ultrasound Physics. St. Louis: Mosby, 1996:46–47

a) Distance accuracy

141

| 09.11.19 - 01:53

Ultrasound Imaging [2] Zagzebski JA. Physics of diagnostic ultrasound. Essentials of Ultrasound Physics. St. Louis: Mosby, 1996:6 [3] Dudley NJ, Gibson NM, Fleckney MJ, Clark PD. The effect of speed of sound in ultrasound test objects on lateral resolution. Ultrasound Med Biol. 2002; 28(11–12):1561–1564 [4] Hangiandreou NJ, Stekel SF, Tradup DJ, Gorny KR, King DM. Four-year experience with a clinical ultrasound quality control program. Ultrasound Med Biol. 2011; 37(8):1350–1357 [5] Powis RL, Moore GW. The silent revolution: catching up with the contemporary composite transducer. J Diagn Med Sonogr. 2004; 20:395–405 [6] American College of Radiology (ACR). Ultrasound accreditation program requirements: http://www.acraccreditation.

142

org/~/media/ACRAccreditation/Documents/Ultrasound/ Requirements.pdf?la=en (accessed 12/26/2018) [7] Avruch L, Cooperberg PL. The ring-down artifact. J Ultrasound Med. 1985; 4(1):21–28 [8] Anvari A, Forsberg F, Samir AE. A primer on the physical principles of tissue harmonic imaging. Radiographics. 2015; 35 (7):1955–1964 [9] DICOM PS3. 14 2018e – Grayscale Standard Display Function, NEMA. http://dicom.nema.org/medical/dicom/current/output/html/part14.html (accessed 12/31/2018)

| 09.11.19 - 01:53

7 Image Processing Jonathon A. Nye and Randahl C. Palmer

Introduction Image processing and display are critical components in the imaging workflow chain. They have tremendous diagnostic utility for disease detectability and

interpretation. In addition, the combination of information from two or more modalities can increase both sensitivity and specificity compared to a single exam. This chapter presents basic image processing concepts that are used in medical imaging.

143

| 09.11.19 - 01:53

Image Processing

7.1 Case 1: Filtering and Edge Enhancement 7.1.1 Background Presentation of an anteroposterior (AP) chest radiograph.

7.1.2 Findings ●

Presentation of post image filtering by convolution with kernels designed to extract features of different frequencies.

●

Kernels are commonly used to lower noise or enhance edge information.

7.1.3 Discussion Filtering is one of the most basic imaging processing steps to enhance image contrast. It can be applied either in frequency space through use of the Fourier transform or in image space through use of the convolution process.1 In image space, filtering involves construction of a kernel that is moved across the image. All pixels within the kernel are averaged and that average is placed in

a new image. A common kernel is a Gaussian function, which is a low-pass operation that reduces noise (e.g., quantum mottle) and improves visibility of low-contrast features. High-contrast features can be enhanced by adding negative lobes to the kernel. This process improves high-contrast resolution but also increases noise since both of these features are high-frequency components of an image. Kernels are normalized to preserve the scale of the original image. The convolution process is described in ▶ Fig. 7.11 where discrete kernel values are multiplied by the pixel values that fall underneath the function and the sum of these products is placed in a new image. The kernel is then shifted and the process is repeated. ▶ Fig. 7.2 demonstrates the change in contrast of a chest radiograph following Gaussian smoothing and Gaussian–Laplacian edge enhancement.

7.1.4 Resolution Convolution using kernels designed to lower noise or enhance edge information alters contrast and can improve the detectability of anatomical features such as soft-tissue masses or bone fractures.

Fig. 7.1 1D example of a convolution of an edge (open circles) with a Gaussian kernel. Data are rounded to the nearest whole number for easier display. The plot shows the original data including the same information following a Gaussian smoothing operation (solid line) and a hybrid Gaussian–Laplacian edge enhancement operation (dashed line).

144

| 09.11.19 - 01:53

7.1 Case 1: Filtering and Edge Enhancement

Fig. 7.2 (a) Chest X-ray without application of a postprocessing convolution filter. (b) Convolved with a 10-pixel fullwidth-at-half-maximum Gaussian filter. (c) Convolved with a 10-pixel full-width-at-half-maximum Gaussian–Laplacian edge-enhancement filter.

145

| 09.11.19 - 01:53

Image Processing

7.2 Case 2: Maximum Intensity Projection 7.2.1 Background ●

Whole-body [18F] fluorodeoxyglucose ([18F] FDG) positron emission tomography (PET) of a patient with extensive disease.

●

The image volume is processed with a raytracing technique called maximum intensity projection (MIP), to highlight hyper-metabolic activity throughout the volume (▶ Fig. 7.3).

7.2.2 Findings ●

Compared to a standard coronal slice, the MIP image permits visualization at depth within a 3D volume on a 2D display.

●

The MIP volume can be rotated to improve visualization of hot lesions that lie along the same ray path but in different planes.

7.2.3 Discussion MIP is a simple but powerful processing tool used to visualize a 3D volumetric dataset. The most straightforward implementation of the method involves using a ray-tracing technique that projects the maximum pixel value along parallel paths within a 3D volume on to a 2D

viewing plane (▶ Fig. 7.4). The volume can be oriented in any direction or rotated incrementally after each ray tracing to produce many 2D projections that can be played back as a movie. The MIP process gives the perception of looking through a volume as opposed to the conventional methods of paging through slices. It is used extensively in nuclear medicine tomography (e.g., PET/SPECT) to quickly identify hot spots in a large 3D volume. Two hot areas of interest that lie along the same ray path can mask one another, therefore rotating the volume can reveal these superimposed areas.

7.2.4 Resolution MIP is a simple yet powerful visual rendering method that gives information about an entire 3D volume without the need to scroll through slices. The method is completely automated, commonly displayed in the coronal orientation, and may be oriented in any direction. More sophisticated implementations can be performed where pixels along the visualization vector are weighted by or limited to a certain depth within the 3D volume. This process is used, in combination with other processing techniques, to construct synthetic 2D mammograms from breast tomosynthesis data. Other common uses of MIP include visualization of vascular structures in computed tomography angiography and magnetic resonance (MR) timeof-flight.2,3,4

Fig. 7.3 (a) Coronal slice of a [18F] FDG whole-body PET/CT showing extensive disease. (b) Maximum intensity projection processed volume in the coronal orientation along the anteriorposterior direction. Note the visualization of lesions at deeper slices that are out of plane in (a).

146

| 09.11.19 - 01:53

7.2 Case 2: Maximum Intensity Projection

Fig. 7.4 (a) Illustration of the maximum intensity projection (MIP) process of ray tracing a 3D volume to create a 2D MIP image. (b) A ray path through pixels finds the maximum value and displays that value on the workstation.

147

| 09.11.19 - 01:53

Image Processing

7.3 Case 3: Fused Image Display of Multiple Modalities

●

Data fusion allows for better identification of anatomical boundaries of disease and visual correlation with changes in metabolism.

7.3.1 Background ●

Independently collected MR and PET images were acquired, registered, and displayed as a single image (▶ Fig. 7.5).

●

The T2-weighted fluid-attenuation inversion recovery (FLAIR) MRI, presented in grayscale, provides structural details that can be visually correlated with functional metabolic information provided by the [18F] FDG brain PET image.

7.3.2 Findings ●

Limited or poor anatomical detail in PET can be augmented by coregistering and fusing these data with a high-resolution anatomical image.

7.3.3 Discussion Display fusion, or color blending, of two or more images is a widely employed technique for displaying and interpreting functional imaging data.5,6 Commonly, the underlying image is a structural modality (e.g., MRI, CT) displayed in grayscale and the overlying image is a functional modality (e.g., PET, SPECT, MRS) displayed in a false color scale. The fusion process is commonly called alpha blending, where images are converted to a 24-bit color image (e.g., red, green, and blue channels) and a transparency value (alpha) is assigned to the blended image. Monitors shipped with common desktop computers are 24-bit, having 3 channels each with 256 shades of color. The displayed image is then a combination of two color scales, grayscale

Fig. 7.5 An example of the alpha blending technique using an magnetic resonance image in grayscale and positron emission tomography image in rainbow color.

148

| 09.11.19 - 01:53

7.3 Case 3: Fused Image Display of Multiple Modalities and false color, that gives the perception of transparency depending on the choice of the alpha value.

7.3.4 Resolution The example in ▶ Fig. 7.5 demonstrates the alpha blending image fusion technique using a linear combination of display scales from two images.

More sophisticated blending techniques are available that can preserve certain features, for example, thresholding the PET image to display only standardized uptake values above a predefined value. Image fusion has enabled hybrid imaging to grow into a powerful diagnostic tool increasing both sensitivity and specificity of interpretation compared to viewing one of the two modalities alone.7

149

| 09.11.19 - 01:53

Image Processing

7.4 Case 4: Multimodality Image Registration

●

Spatial coregistration can be accomplished using automated routines based on similarity metrics including entropy and mutual information.

7.4.1 Background ●

[18F] FDG and T2-weighted FLAIR brain images

7.4.3 Background

were acquired in a patient with epilepsy. ●

Images were collected on different instruments,

Patient data collected from different scanners can be coregistered to correlate voxel intensities of function (e.g., PET, SPECT) with areas of morphology (CT, MRI). For imaging data collected on stand-alone systems (nonhybrid), a software solution is needed. There are a large number of coregistration similarity metrics for determining whether two images are aligned well. Two common approaches include use of external surface landmarks and correlation of voxel intensity but these methods are either cumbersome to

therefore there is a need to spatially align them to the same reference orientation.

7.4.2 Findings ●

Fused display of the images show misalignment because these data were acquired on standalone systems (▶ Fig. 7.6a).

Fig. 7.6 Examples of misregistered (a, c) and registered (b, d) images of positron emission tomography with magnetic resonance imaging (a, b) and computed tomography (c, d). The second column are the joint histograms of pixel intensities between the respective images. The third column lists the calculated joint entropy and mutual information of the images in their orientation as displayed in the first column. Note that lower joint entropy and higher mutual information indicate better spatial alignment.

150

| 09.11.19 - 01:53

7.4 Case 4: Multimodality Image Registration implement or present difficulties when voxels of two different modalities are weakly correlated, respectively. A popular similarity metric for intermodality image registration is the one based on entropy.8,9 Entropy is a measure of dispersion in the probability distribution. The probably distribution between two images can be represented by plotting a 2D histogram. The joint histogram counts how many times a grayscale value correspondence occurs; the brighter the pixel in the joint histogram, the larger the correspondence frequency. Entropy can be thought of as a measure of the histogram dispersion. Minimizing the dispersion of the joint histogram occurs when the joint entropy between two image modalities is minimized. A common implementation of entropy is maximization of mutual information (MI) given as Entropy ðMRIÞ þ Entropy ðPETÞ − Entropy ðMRI þ PETÞ MI is the most commonly used similarity metric and is considered highly robust and reliable. ▶ Fig. 7.6 demonstrates the differences in sensitivity between entropy and MI when comparing misregistered and registered PET images with either CT or MRI.

7.4.4 Resolution

by maximizing a similarity measure. Similarity measures based on entropy have prevailed, as they are highly reliable when aligning two images from different modalities.

References [1] Brody WR. In: Bankman IN, ed. Handbook of Medical Image Processing and Analysis. 2nd ed. Burlington: Academic Press; 2009 [2] Prokop M, Shin HO, Schanz A, Schaefer-Prokop CM. Use of maximum intensity projections in CT angiography: a basic review. Radiographics. 1997; 17(2):433–451 [3] Saloner D. The AAPM/RSNA physics tutorial for residents. An introduction to MR angiography. Radiographics. 1995; 15(2): 453–465 [4] Hofman MS, Hicks RJ. How we read oncologic FDG PET/CT. Cancer Imaging. 2016; 16(1):35 [5] Bican J, Janeba D, Táborská K, Veselý J. Image overlay using alpha-blending technique. Nucl Med Rev Cent East Eur. 2002; 5(1):53 [6] Watson CC, Townsend DW, Bendriem B. PET/CT systems. In: Wernick MN, Aarsvold JN, eds. Emission Tomography. San Diego: Academic Press; 2004:195–212 [7] Hany TF, Steinert HC, Goerres GW, Buck A, von Schulthess GK. PET diagnostic accuracy: improvement with in-line PETCT system: initial results. Radiology. 2002; 225(2):575–581 [8] Viergever MA, Maintz JBA, Klein S, Murphy K, Staring M, Pluim JPW. A survey of medical image registration—under review. Med Image Anal. 2016; 33:140–144 [9] Oliveira FP, Tavares JM. Medical image registration: a review. Comput Methods Biomech Biomed Engin. 2014; 17(2):73–93

Image registration is an optimization problem where the goal is to match corresponding features

151

| 09.11.19 - 01:53

Answer Key Chapter 1 Fluoroscopy 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

a b e c b a d a a b a c b c c b b a d c

Chapter 2 Mammography 1. 2. 3. 4. 5. 6. 7.

8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.

152

a b c b b b b Due to the limited angle scan, depth resolution in digital tomosynthesis is much lower than the in-plane resolution. c d b b b e e b d b d c d c b

Chapter 3 Computed Tomography 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

a b a a d b c d b a a b b b a a d a c b

Chapter 4 Magnetic Resonance Imaging 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

b c b b b a c c d c a a b c b c d b d

| 09.11.19 - 01:53

Answer Key

Chapter 5 Nuclear Medicine 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

b a a b a c b b d a a c b d b a a b c b

5.

6.

7.

Chapter 6 Ultrasound Imaging 1. a If the true speed of sound propagation is slower than 1,540 m/s, it takes longer time to travel; thus the distance will be overestimated. [Zagzebski JA. Pulse-echo ultrasound instrumentation. Essentials of Ultrasound Physics. St. Louis: Mosby, 1996:46-47] 2. c The speed of sound is dependent on the properties of the medium such as its density and compressibility. [Zagzebski JA. Physics of diagnostic ultrasound. Essentials of Ultrasound Physics. St. Louis: Mosby, 1996:6] 3. b Curved array produces acoustic beams diverging as the depth increases and its transducer surface has a wide footprint. [Zagzebski JA. Physics of diagnostic ultrasound. Essentials of Ultrasound Physics. St. Louis: Mosby; 1996:34-3] 4. a This image has the lowest depth of penetration due to high attenuation at high frequency.

8.

9.

10.

11.

12.

[Zagzebski JA. Physics of diagnostic ultrasound. Essentials of Ultrasound Physics. St. Louis: Mosby; 1996:26-27] c The nonuniformity near surface is typically due to dead transducer elements. [ACR technical standard for diagnostic medical physics performance monitoring of real time ultrasound equipment, Revised 2016] c Image uniformity is an effective phantom test for revealing transducer element dropouts. [Hangiandreou NJ, Stekel SF, Tradup DJ, Gorny KR, King DM. Four-year experience with a clinical ultrasound quality control program. Ultrasound Med Biol. 2011; 37 (8):1350-1357] d Only the increase of the transmit power will increase the depth of penetration. The increase of the dynamic range or the overall gain will not affect the maximum depth of penetration. Increasing the transducer frequency will reduce penetration. [Zagzebski JA. Pulse-echo ultrasound instrumentation. Essentials of Ultrasound Physics. St. Louis: Mosby; 1996] a The horizontal band of darkness may be caused by insufficient amplification at the specific depth and can be adjusted by sliding the TGC knob that controls the amplification at the corresponding depth. [Zagzebski JA. Pulse-echo ultrasound instrumentation. Essentials of Ultrasound Physics. St. Louis: Mosby; 1996] b [Zagzebski JA. Pulse-echo ultrasound instrumentation. Essentials of Ultrasound Physics. St. Louis: Mosby; 1996] b [Zagzebski JA. Pulse-echo ultrasound instrumentation. Essentials of Ultrasound Physics. St. Louis: Mosby; 1996] d [Hangiandreou NJ, O'Brien RT, Zagzebski JA, Delaney FA. Ultrasound corner: range ambiguity artifact. Vet Radiol Ultrasound. 2001; 42(6):542-545] b

153

| 09.11.19 - 01:53

Answer Key

13.

14.

15.

16.

154

[Hangiandreou NJ, O'Brien RT, Zagzebski JA, Delaney FA. Ultrasound corner: range ambiguity artifact. Vet Radiol Ultrasound. 2001; 42(6):542-545] a [Zagzebski JA. Pulse-echo ultrasound instrumentation. Essentials of Ultrasound Physics. St. Louis: Mosby; 1996] b [Zagzebski JA. Pulse-echo ultrasound instrumentation. Essentials of Ultrasound Physics. St. Louis: Mosby; 1996] b [Anvari A, Forsberg F, Samir AE. A primer on the physical principles of tissue harmonic imaging. Radiographics. 2015; 35 (7):1955-1964] c [Anvari A, Forsberg F, Samir AE. A primer on the physical principles of tissue harmonic imaging. Radiographics. 2015; 35 (7):1955-1964]

17. a [On-Line Report No AAPM. 03, Assessment of Display Performance for Medical Imaging Systems, American Association of Physicists in Medicine, 2005. https://www.aapm.org/ pubs/reports/OR_03.pdf (accessed on 12/ 31/2018)] 18. b [American College of Radiology (ACR). Ultrasound accreditation program requirements: http://www.acraccreditation.org/ ~/media/ACRAccreditation/Documents/ Ultrasound/Requirements.pdf?la=en (accessed 12/26/2018)] 19. a [Zagzebski JA. Pulse-echo ultrasound instrumentation. Essentials of Ultrasound Physics. St. Louis: Mosby; 1996] 20. c [Zagzebski JA. Pulse-echo ultrasound instrumentation. Essentials of Ultrasound Physics. St. Louis: Mosby; 1996]

| 09.11.19 - 01:53

Index A abnormal dark fluid seen in bladder of an axial single-Shot T2-weighted sequence – background 76 – discussion 76 – findings 76 – resolution 78 antiscatter grids – background 8 – discussion 8 – findings 8 – resolution 9 array transducers and sound frequency – background 120 – discussion 121 – findings 120 artifact due to detector row dropout – background 35 – discussion 35 – findings 35 – resolution 35 artifact due to imperfection in compression paddle – background 37 – discussion 37 – findings 37 – resolution 37 attenuation correction in PET – background 99 – discussion 99 – findings 99 – resolution 100

B beam hardening artifact – background 52 – discussion 52 – findings 52 – resolution 52 bremsstrahlung imaging of 90Y microspheres liver embolization – background 110 – discussion 110 – findings 110 – resolution 110

C collimation – background 6 – discussion 6 – findings 6 – resolution 6 computed tomography – beam hardening artifact –– background 52 –– discussion 52 –– findings 52 –– resolution 52

– displayed volume CT dose index –– background 51 –– discussion 51 –– findings 51 –– resolution 51 – image quality variation with reconstructed slice thickness –– background 47 –– discussion 47 –– findings 47 –– resolution 47 – image quality variation with reconstruction filter –– background 49 –– discussion 49 –– findings 49 –– resolution 49 – kV selection on image quality and dose, effect of –– background 46 –– discussion 46 –– findings 46 –– resolution 46 – metal artifact –– background 54 –– discussion 54 –– findings 54 –– resolution 54 – motion artifact –– background 55 –– discussion 55 –– findings 55 –– resolution 55 – partial volume artifact –– background 53 –– discussion 53 –– findings 53 –– resolution 53 – patient size on CT number accuracy, effect of –– background 45 –– discussion 45 –– findings 45 –– resolution 45 – ring artifact –– background 44 –– discussion 44 –– findings 44 –– resolution 44 cone beam computed tomography – background 20 – discussion 20 – findings 20 – resolution 20 correct acquisition image matrix size – background 106 – discussion 106 – findings 106 – resolution 107

CT fluoroscopy – – – –

background 12 discussion 12 findings 12 resolution 12

D dark etching appears at boundary of fat and soft-tissue layers – background 69 – discussion 69 – findings 69 – resolution 69 degraded image quality from improper collimator – background 112 – discussion 112 – findings 112 – resolution 112 degraded resolution of whole-body planar 99mTc methylene diphosphonate image – background 91 – discussion 91 – findings 91 – resolution 91 digital breast tomosynthesis – background 30 – discussion 30 – findings 31 – resolution 31 digital subtraction angiography and motion artifacts – background 14 – discussion 14 – findings 14 – resolution 15 discrete image ghosts on abdominal imaging, appearance of – background 60 – discussion 60 – findings 60 – resolution 61 displayed volume CT dose index – background 51 – discussion 51 – findings 51 – resolution 51 doppler ultrasound aliasing – background 135 – discussion 135 – findings 135 – resolution 135

E EMI artifact due to LVAD device – background 39 – discussion 39

155

| 09.11.19 - 01:53

Index – findings 39 – resolution 39 equalization filters – background 18 – discussion 18 – findings 18 – resolution 18 extra field-of-view anatomy on inferior portion of sagittal 3D T2-weighted acquisition of spine, appearance of – background 65 – discussion 65 – finding 65 – resolution 66

F fat-suppressed sequences, application of – background 71 – discussion 71 – findings 71 – resolution 72 filtering and edge enhancement – background 144 – discussion 144 – findings 144 – resolution 144 fluoroscopy – antiscatter grids –– background 8 –– discussion 8 –– findings 8 –– resolution 8 – collimation –– background 6 –– discussion 6 –– findings 6 –– resolution 7 – cone beam computed tomography –– background 20 –– discussion 20 –– findings 20 –– resolution 21 – CT fluoroscopy –– background 12 –– discussion 12 –– findings 12 –– resolution 12 – digital subtraction angiography and motion artifacts –– background 14 –– discussion 14 –– findings 14 –– resolution 15 – equalization filters –– background 18 –– discussion 18 –– findings 18 –– resolution 18 – modes and dose –– background 16 –– discussion 16 –– findings 16 –– resolution 16 – patient shielding

156

–– background 10 –– discussion 10 –– findings 10 –– resolution 11 – reference air kerma and skin dose –– background 4 –– discussion 4 –– findings 4 –– resolution 5 – SID, ABC, and radiation output 12 –– background 2 –– discussion 2 –– findings 2 –– resolution 2 focal spot size selection in magnification views – background 27 – discussion 27 – findings 27 – resolution 28 fused image display of multiple modalities – background 148 – discussion 148 – findings 148 – resolution 149

H harmonic imaging – background 132 – discussion 132 – findings 132 – resolution 132 hyperintensity appears bilaterally at the level of the internal auditory canal on diffusion-weighted MRI, affecting visualization of surrounding structures – background 62 – discussion 62 – findings 62 – resolution 63

I image post-processing on appearance – background 32 – discussion 32 – findings 34 – resolution 34 image processing – filtering and edge enhancement –– background 144 –– discussion 144 –– findings 144 –– resolution 144 – fused image display of multiple modalities –– background 148 –– discussion 148 –– findings 148 –– resolution 149 – maximum intensity projection –– background 146 –– discussion 146 –– findings 146

–– resolution 146 – multimodality image registration –– background 150 –– discussion 150 –– findings 150 –– resolution 151 image quality variation with reconstructed slice thickness – background 47 – discussion 47 – findings 47 – resolution 47 image quality variation with reconstruction filter – background 49 – discussion 49 – findings 49 – resolution 49 image smoothing – background 104 – discussion 104 – findings 104 – resolution 105 iterative reconstruction and choosing the number of iterations and subsets – background 101 – discussion 101 – findings 101 – resolution 101

K kV selection on image quality and dose, effect of – background 46 – discussion 46 – findings 46 – resolution 46

M magnetic resonance imaging – abnormal dark fluid seen in bladder of an axial single-Shot T2-weighted sequence, but not on location-matched 3D T2 acquistion –– background 76 –– discussion 76 –– findings 76 –– resolution 78 – dark etching appears at boundary of fat and soft-tissue layers 67 –– background 69 –– discussion 69 –– findings 69 –– resolution 69 – appearance of discrete image ghosts on abdominal imaging 60 –– background 60 –– discussion 60 –– findings 60 –– resolution 61 – extra field-of-view anatomy on inferior portion of sagittal 3D T2-weighted acquisition of spine, appearance of 65

| 09.11.19 - 01:53

Index –– background 65 –– discussion 65 –– finding 65 –– resolution 66 – fat-suppressed sequences, application of 71 –– background 71 –– discussion 71 –– findings 71 –– resolution 72 – hyperintensity appears bilaterally at level of internal auditory canal on diffusion-weighted MRI, affecting visualization of surrounding structures –– background 62 –– discussion 62 –– findings 62 –– resolution 63 – post contrast T1-weighted gradient echo reveals patchy enhancement in anterior septal wall 79 –– background 79 –– discussion 79 –– findings 79 –– resolution 81 – precontrast, axial 3D T1-weighted gradient echo with fat suppressions 67 –– background 67 –– discussion 67 –– findings 67 –– resolution 68 – signal-to-noise variation across FOV, creating nondiagnostic image quality –– background 82 –– discussion 82 –– findings 82 –– resolution 84 – T1-weighted gradient echo of abdomen shows marked artifact medially on both coronal and axial FOV, obscuring visualization of soft tissues –– discussion 74 –– Fbackground 74 –– findings 74 –– resolution 75 magnification imaging – background 26 – discussion 26 – findings 26 – resolution 26 mammography – artifact due to detector row dropout 35 –– background 35 –– discussion 35 –– findings 35 –– resolution 35 – artifact due to imperfection in compression paddle 37 –– background 37 –– discussion 37 –– findings 37 –– resolution 37 – EMI artifact due to LVAD device –– background 39 –– discussion 39 –– findings 39

–– resolution 39 – equations 42 – focal spot size selection in magnification views 27 –– background 27 –– discussion 27 –– findings 27 –– resolution 28 – image post-processing on appearance 32 –– background 32 –– discussion 32 –– findings 34 –– resolution 34 – magnification imaging –– background 26 –– discussion 26 –– findings 26 –– resolution 26 – microcalcification-like appearance caused by detector artifact –– background 36 –– discussion 36 –– findings 36 –– resolution 36 – patient motion causing blurred parenchymal structure –– background 38 –– discussion 38 –– findings 38 –– resolution 38 – x-ray acquisition technique factors –– background 29 –– discussion 29 –– findings 29 –– resolution 30 maximum intensity projection – background 146 – discussion 146 – findings 146 – resolution 146 metal artifact – background 54 – discussion 54 – findings 54 – resolution 54 microcalcification-like appearance caused by detector artifact – background 36 – discussion 36 – findings 36 – resolution 36 modes and dose – background 16 – discussion 16 – findings 16 – resolution 16 motion artifact – background 55 – discussion 55 – findings 55 – resolution 55 multimodality image registration – background 150 – findings 150 – resolution 151

N nonuniformity (array transducer element dropouts) – background 122 – discussion 122 – findings 122 – resolution 122 nuclear medicine – attenuation correction in PET –– background 99 –– discussion 99 –– findings 99 –– resolution 99 – bremsstrahlung imaging of 90Y microspheres liver embolization –– background 110 –– discussion 110 –– findings 110 –– resolution 110 – correct acquisition image matrix size –– background 106 –– discussion 106 –– findings 106 –– resolution 107 – degraded image quality from improper collimator –– background 112 –– discussion 112 –– findings 112 –– resolution 112 – degraded resolution of whole-body planar 99mTc methylene diphosphonate image –– background 91 –– discussion 91 –– findings 91 –– resolution 92 – image smoothing –– background 104 –– discussion 104 –– findings 104 –– resolution 105 – iterative reconstruction and choosing the number of iterations and subsets –– background 101 –– discussion 101 –– findings 101 –– resolution 101 – patient motion in myocardial perfusion imaging –– background 108 –– discussion 108 –– findings 108 –– resolution 108 – positron range on image quality and resolution –– background 94 –– findings 94 –– discussion 94 –– resolution 96 – standardized uptake value in positron emission tomography –– background 97 –– discussion 97

157

| 09.11.19 - 01:53

Index –– findings 97 –– resolution 98

P partial volume artifact – background 53 – discussion 53 – findings 53 – resolution 53 patient motion causing blurred parenchymal structure – background 38 – discussion 38 – findings 38 – resolution 38 patient motion in myocardial perfusion imaging – background 108 – discussion 108 – findings 108 – resolution 108 patient shielding – background 10 – discussion 10 – findings 10 – resolution 11 patient size on CT number accuracy, effect of – background 45 – discussion 45 – findings 45 – resolution 45 positron range on image quality and resolution – background 94 – findings 94 – discussion 94 – resolution 96 post contrast T1-weighted gradient echo reveals patchy enhancement in anterior septal wall – background 79 – discussion 79 – findings 79 – resolution 81 precontrast, axial 3D T1-weighted gradient echo with fat suppressions – background 67 – discussion 67 – findings 67 – resolution 68 pulse-echo imaging acquisition – background 124 – findings 124 – discussion 124 – resolution 126 pulse-echo imaging principle and speed of sound propagation – background 118 – discussion 118 – findings 118 – resolution 118

158

R range ambiguity in B-mode – background 129 – findings 129 – discussion 129 – resolution 129 reference air kerma and skin dose – background 4 – discussion 4 – findings 4 – resolution 5 reflection (boundary conditions)-reverberation artifacts – background 127 – discussion 127 – findings 127 – resolution 128 ring artifact – background 44 – discussion 44 – findings 44 – resolution 44

S shadowing and enhancement – background 130 – findings 130 – discussion 130 – resolution 130 SID, ABC, and radiation output – background 2 – discussion 2 – findings 2 – resolution 2 signal-to-noise variation across FOV, creating nondiagnostic image quality – background 82 – discussion 82 – findings 82 – resolution 84 standardized uptake value in positron emission tomography – background 97 – discussion 97 – findings 97 – resolution 98

T T1-weighted gradient echo of abdomen shows marked artifact medially on both coronal and axial FOV, obscuring visualization of soft tissues – background 74 – discussion 74 – findings 74 – resolution 75

U ultrasound image display on scanners and in reading rooms – background 133

– findings 133 – discussion 133 – resolution 134 ultrasound imaging – array transducers and sound frequency –– background 120 –– discussion 121 –– findings 120 –– resolution 121 – doppler ultrasound aliasing –– background 135 –– discussion 135 –– findings 135 –– resolution 135 – harmonic imaging –– background 132 –– findings 132 –– discussion 132 –– resolution 132 – nonuniformity (array transducer element dropouts) –– background 122 –– discussion 122 –– findings 122 –– resolution 123 – pulse-echo imaging acquisition –– background 124 –– findings 124 –– discussion 124 –– resolution 126 – pulse-echo imaging principle and speed of sound propagation –– background 118 –– discussion 118 –– findings 118 –– resolution 118 – range ambiguity in B-mode –– background 129 –– findings 129 –– discussion 129 –– resolution 129 – reflection (boundary conditions)-reverberation artifacts –– background 127 –– discussion 127 –– findings 127 –– resolution 128 – shadowing and enhancement –– background 130 –– findings 130 –– discussion 130 –– resolution 130 – ultrasound image display on scanners and in reading rooms –– background 133 –– findings 133 –– discussion 133 –– resolution 134

X X -ray acquisition technique factors – background 29 – discussion 29 – findings 29 – resolution 30