P1: OTA/XYZ P2: ABC
fm JWBS035-Haidekker August 31, 2010 19:57 Printer Name: Yet to Come
ADVANCED BIOMEDICAL
IMAGE ANALYSIS
P1: OTA/XYZ P2: ABC
fm JWBS035-Haidekker August 31, 2010 19:57 Printer Name: Yet to Come
ADVANCED BIOMEDICAL
IMAGE ANALYSIS
MARK A. HAIDEKKER
A JOHN WILEY & SONS, INC., PUBLICATION
P1: OTA/XYZ P2: ABC
fm JWBS035-Haidekker August 31, 2010 19:57 Printer Name: Yet to Come
Copyright
C 2011 by John Wiley & Sons, Inc. All rights reserved.
Published by John Wiley & Sons, Inc., Hoboken, New Jersey.
Published simultaneously in Canada.
No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or
by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as
permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior
written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to
the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400,
fax (978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher for permission
should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken,
NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permission.
Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in
preparing this book, they make no representations or warranties with respect to the accuracy or
completeness of the contents of this book and specifically disclaim any implied warranties of
merchantability or fitness for a particular purpose. No warranty may be created or extended by sales
representatives or written sales materials. The advice and strategies contained herein may not be suitable
for your situation. You should consult with a professional where appropriate. Neither the publisher nor
author shall be liable for any loss of profit or any other commercial damages, including but not limited to
special, incidental, consequential, or other damages.
For general information on our other products and services or for technical support, please contact our
Customer Care Department within the United States at (800) 762-2974, outside the United States at
(317) 572-3993 or fax (317) 572-4002.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may
not be available in electronic formats. For more information about Wiley products, visit our web site at
www.wiley.com.
Library of Congress Cataloging-in-Publication Data:
Haidekker, Mark A.
Advanced biomedical image analysis / Mark A. Haidekker
Includes bibliographical references and index.
ISBN 978-0-470-62458-6
Printed in Singapore
10 9 8 7 6 5 4 3 2 1
P1: OTA/XYZ P2: ABC
fm JWBS035-Haidekker August 31, 2010 19:57 Printer Name: Yet to Come
CONTENTS
Preface ix
1 Image Analysis: A Perspective 1
1.1 Main Biomedical Imaging Modalities, 3
1.2 Biomedical Image Analysis, 7
1.3 Current Trends in Biomedical Imaging, 12
1.4 About This Book, 15
References, 17
2 Survey of Fundamental Image Processing Operators 23
2.1 Statistical Image Description, 24
2.2 Brightness and Contrast Manipulation, 28
2.3 Image Enhancement and Restoration, 29
2.4 Intensity-Based Segmentation (Thresholding), 42
2.5 Multidimensional Thresholding, 50
2.6 Image Calculations, 54
2.7 Binary Image Processing, 58
2.8 Biomedical Examples, 63
References, 68
3 Image Processing in the Frequency Domain 70
3.1 The Fourier Transform, 71
3.2 Fourier-Based Filtering, 82
v
P1: OTA/XYZ P2: ABC
fm JWBS035-Haidekker August 31, 2010 19:57 Printer Name: Yet to Come
vi CONTENTS
3.3 Other Integral Transforms: The Discrete Cosine Transform
and the Hartley Transform, 91
3.4 Biomedical Examples, 94
References, 100
4 The Wavelet Transform and Wavelet-Based Filtering 103
4.1 One-Dimensional Discrete Wavelet Transform, 106
4.2 Two-Dimensional Discrete Wavelet Transform, 112
4.3 Wavelet-Based Filtering, 116
4.4 Comparison of Frequency-Domain Analysis to Wavelet
Analysis, 128
4.5 Biomedical Examples, 130
References, 135
5 Adaptive Filtering 138
5.1 Adaptive Noise Reduction, 139
5.2 Adaptive Filters in the Frequency Domain: Adaptive Wiener
Filters, 155
5.3 Segmentation with Local Adaptive Thresholds and Related
Methods, 157
5.4 Biomedical Examples, 164
References, 170
6 Deformable Models and Active Contours 173
6.1 Two-Dimensional Active Contours (Snakes), 180
6.2 Three-Dimensional Active Contours, 193
6.3 Live-Wire Techniques, 197
6.4 Biomedical Examples, 205
References, 209
7 The Hough Transform 211
7.1 Detecting Lines and Edges with the Hough Transform, 213
7.2 Detection of Circles and Ellipses with the Hough Transform, 219
7.3 Generalized Hough Transform, 223
7.4 Randomized Hough Transform, 226
7.5 Biomedical Examples, 231
References, 234
8 Texture Analysis 236
8.1 Statistical Texture Classification, 238
8.2 Texture Classification with Local Neighborhood Methods, 242
8.3 Frequency-Domain Methods for Texture Classification, 254
P1: OTA/XYZ P2: ABC
fm JWBS035-Haidekker August 31, 2010 19:57 Printer Name: Yet to Come
CONTENTS vii
8.4 Run Lengths, 257
8.5 Other Classification Methods, 263
8.6 Biomedical Examples, 265
References, 273
9 Shape Analysis 276
9.1 Cluster Labeling, 278
9.2 Spatial-Domain Shape Metrics, 279
9.3 Statistical Moment Invariants, 285
9.4 Chain Codes, 287
9.5 Fourier Descriptors, 291
9.6 Topological Analysis, 295
9.7 Biomedical Examples, 301
References, 307
10 Fractal Approaches to Image Analysis 310
10.1 Self-Similarity and the Fractal Dimension, 311
10.2 Estimation Techniques for the Fractal Dimension in
Binary Images, 319
10.3 Estimation Techniques for the Fractal Dimension in
Gray-Scale Images, 327
10.4 Fractal Dimension in the Frequency Domain, 331
10.5 Local Hölder Exponent, 337
10.6 Biomedical Examples, 340
References, 345
11 Image Registration 350
11.1 Linear Spatial Transformations, 352
11.2 Nonlinear Transformations, 355
11.3 Registration Quality Metrics, 360
11.4 Interpolation Methods for Image Registration, 371
11.5 Biomedical Examples, 379
References, 382
12 Image Storage, Transport, and Compression 386
12.1 Image Archiving, DICOM, and PACS, 389
12.2 Lossless Image Compression, 392
12.3 Lossy Image Compression, 400
12.4 Biomedical Examples, 408
References, 411
13 Image Visualization 413
13.1 Gray-Scale Image Visualization, 413
P1: OTA/XYZ P2: ABC
fm JWBS035-Haidekker August 31, 2010 19:57 Printer Name: Yet to Come
viii CONTENTS
13.2 Color Representation of Gray-Scale Images, 416
13.3 Contour Lines, 422
13.4 Surface Rendering, 422
13.5 Volume Visualization, 427
13.6 Interactive Three-Dimensional Rendering and Animation, 433
13.7 Biomedical Examples, 434
References, 438
14 Image Analysis and Visualization Software 441
14.1 Image Processing Software: An Overview, 443
14.2 ImageJ, 447
14.3 Examples of Image Processing Programs, 452
14.4 Crystal Image, 456
14.5 OpenDX, 461
14.6 Wavelet-Related Software, 466
14.7 Algorithm Implementation, 466
References, 473
Appendix A: Image Analysis with Crystal Image 475
Appendix B: Software on DVD 497
Index 499
P1: OTA/XYZ P2: ABC
fm JWBS035-Haidekker August 31, 2010 19:57 Printer Name: Yet to Come
PREFACE
Medical imaging is one of the great revolutions in medicine. Traditionally, explo-
rative surgery had to be performed to look inside a patient’s body, even to perform a
diagnosis. Slightly more than a century ago, x-rays were discovered. With it came the
ability to look inside the body without surgery. X-ray imaging was rapidly adopted
in medical centers worldwide. A new medical subject area was created—radiology.
For decades, the radiologist had a basic set of tools, the x-ray tube, a fluoroscope, a
film cassette with an image intensifier screen, and a light box. For decades, progress
was incremental. X-ray tubes were improved, film and intensifier were made more
sensitive, radiation exposure was reduced, and contrast agents were introduced and
improved. It took a second, independent revolution to propel biomedical imaging to
today’s level: the invention of the programmable computer and its subsequent minia-
turization. The availability of powerful digital data processing hardware and newly
developed image processing methods paved the way for new imaging modalities:
computed tomography, magnetic resonance imaging, ultrasound imaging, and func-
tional imaging. These new imaging modalities had in common that computer-based
data processing was required for image formation. Medical imaging experienced a
second wave of rapid progress near the end of the twentieth century when tomography
methods were developed and improved. Tomography means imaging by sections, and
the origin of the word lies in the Greek óo for “to cut” and ␥␣
´ for “to write.”
With the availability of imaging modalities that produced three-dimensional recon-
structions of a patient’s body came the need for computerized image processing and
computerized image visualization. The expertise of the radiologist in interpreting an
image played–and still plays—a major role, but more and more tasks could be given
to the computer, and the interpretation of computer-processed images became eas-
ier, more objective, and more accurate. Concurrently, scientists became interested in
ix
P1: OTA/XYZ P2: ABC
fm JWBS035-Haidekker August 31, 2010 19:57 Printer Name: Yet to Come
x PREFACE
programming computers with the “intelligence” to interpret and understand images.
At the same time that computed tomography and magnetic resonance imaging were
being invented, new computer methods for image analysis were being introduced.
The long-term vision is computer-aided radiology. Generalized to nonmedical
fields, we could call it computer-aided image interpretation. After the first major
wave of innovation in the 1970s to 1980s, computer algorithms for image interpreta-
tion have become more sophisticated and more complex. New mathematical methods
emerged, such as, for example, the wavelet transform and set characterization by the
fractal dimension, and were rapidly translated into advanced image analysis methods.
Numerical methods that model physical processes—for example, differential equa-
tions for diffusion or motion—were applied to image processing tasks as diverse as
noise reduction and segmentation. Artificial intelligence models are being used for
computer analysis of high-dimensional feature vectors with the purpose of classify-
ing the underlying pixels. Yet the philosopher’s stone of image processing has not
been discovered: to make a computer interpret an image with the same flexibility and
immunity against artifacts as those of a human observer.
Although this book is well suited as a textbook for graduate-level image processing
classes in the computer sciences and engineering fields, it is intended primarily
as a reference book. The individual chapters are widely independent of the other
chapters, with the exception of Chapters 2 and 3, which provide the foundation of
basic image processing operations. The book is a comprehensive hands-on reference
of topical image processing methods. Hands-on in this context indicates that not
only are the theory, mathematical foundation, and basic description of an image
processing operator provided, but performance features, advantages, and limitations
are also discussed. Furthermore, key algorithms are provided in pseudocode to assist
in implementation. The book aims at making advanced image processing operators
accessible to those readers that have a basic familiarity with image processing. As
such, the book can be seen as a toolbox in which each tool comes with a complete
instruction manual. It is useful for readers who use the tools, because it helps them
understand how the tools work. It is also useful for readers who make the tools, as it
helps them design and optimize tools for their specific needs. And it is intended as a
stepping-stone for those members of the imaging community who are reaching out
to develop the next generation of tools.
The application focus of this book is biomedical. However, the same image pro-
cessing and analysis principles also apply in many other fields. In the environmental
and geological sciences, oceanography, soil sciences, forensic sciences, and anthro-
pology, image analysis plays an important role. Satellite imagery and astrophotogra-
phy, for example, pose the same image processing challenges as those of a magnetic
resonance scan: the need to reduce noise, to emphasize the details of interest, and
to make the important objects in the image accessible for subjective evaluation or
objective measurement.
Acknowledgments
A number of people helped in a significant way to make this book a reality. First and
foremost, I want to give my special thanks to my acquisition editor, Wayne Yuhasz,
P1: OTA/XYZ P2: ABC
fm JWBS035-Haidekker August 31, 2010 19:57 Printer Name: Yet to Come
PREFACE xi
for insisting on this book, for his confidence, and for his continual support throughout
the writing process. I would also like to thank George J. Telecki, Lucy Hitz, and
Angioline Loredo from John Wiley & Sons for their support during manuscript
preparation and production. I particularly appreciated the help of my research
specialist, Darcy Lichlyter, who spent many hours and after-hours in the library
searching for literature no matter how well hidden it was. Adnan Mustafic, one of
my graduate students, showed his competence, initiative, and skill by preparing the
live DVD that accompanies this book. Professor Paul J. Friedman kindly provided
me with a number of image samples. I also want to express my most sincere thanks
to Professor Geoff Dougherty for reviewing the book manuscript and for providing
numerous helpful and excellent suggestions for improving the book, as well as to
Professor Michael Covington for many additional valuable comments. Last, but not
least, I would like to thank my wife, Silke, who not only encouraged me to write
the book and who endured some 80 weekends of my quasi-absence, but who also
applied her expert skill as an accredited editor in the life sciences to proofread and
copyedit the entire manuscript.
Mark A. Haidekker
