SPIE PRESS Bellingham, Washington USA Library of Congress Cataloging-in-Publication Data Computer-aidedcancerdetectionanddiagnosis:recentadvances/JinshanTangand Sos S. Agaian, editors. p. ; cm. Includes bibliographical references and index. ISBN 978-0-8194-9739-0 I. Tang, Jinshan, editor of compilation. II. Agaian, S. S., editor of compilation. [DNLM: 1. Neoplasms diagnosis. 2. Early Detection of Cancer methods. 3. Image Interpretation, Computer-Assisted methods. QZ 241] RC270 616.99'4075 dc23 2013032266 Published by SPIE The International Society for Optical Engineering P.O. Box 10 Bellingham, Washington 98227-0010 USA Phone: +1 360 676 3290 Fax: +1 360 647 1445 Email: [email protected] Web: http://spie.org Copyright © 2014 Society of Photo-Optical Instrumentation Engineers (SPIE) All rights reserved. No part of this publication may be reproduced or distributed in any form or by any means without written permission of the publisher. Thecontentofthisbookreflectstheworkandthoughtoftheauthor(s).Everyefforthas beenmadetopublishreliableandaccurateinformationherein,butthepublisherisnot responsible for the validity of the information or for any outcomes resulting from reliance thereon. Printed in the United States of America. First printing Table of Contents Preface xi List of Contributors xvii 1 Computer-Aided Detection of Colonic Polyps in CT Colonography 1 Jianhua Yao 1.1 Colonic Polyps and Colon Cancer 1 1.2 CT Colonography 2 1.3 Computer-Aided Detection Using CTC 3 1.3.1 CAD pipeline 4 1.3.2 Colon segmentation 5 1.3.3 Supine–prone registration 6 1.3.4 Colon unfolding 8 1.3.5 Polyp segmentation 9 1.3.6 Polyp characterization and features 11 1.3.7 Machine learning and classification 12 1.3.8 Content-based image retrieval 13 1.3.9 CAD performance 14 1.4 Discussion 16 References 17 2 Preprocessing Tools for Computer-Aided Cancer Imaging Systems 23 Artyom M. Grigoryan and Sos S. Agaian 2.1 Introduction 24 2.1.1 Novel view on image processing 25 2.2 Transform-Based Image Enhancement 28 2.2.1 Quantitative measure of image enhancement 29 2.3 Tensor Representation of the Image 30 2.4 Decomposition by Direction Images 33 2.5 Tensor Transform Method of a-Rooting 37 2.5.1 Effective formula for image enhancement 39 2.5.2 Algorithm of image enhancement by 1D a-rooting 42 2.6 Decomposition by the 2D Paired Transform 42 2.6.1 Fourier transform splitting theorem 44 2.6.2 Complete set of the 2D paired transform 46 v vi TableofContents 2.7 Paired Direction Images 47 2.7.1 Principle of superposition by direction images 49 2.7.2 Paired method of image enhancement 49 2.8 Enhancement by a Series of Direction Images 52 2.9 Compression: Multiresolution Map of the Image 55 2.9.1 A-series linear transformation 56 2.10 Compression by the Tensor Transform 59 2.10.1 Block-tensor-transform lossy image compression 62 2.11 Tensor Transform in Image Cryptography 65 2.12 Conclusion 69 References 70 3 MultimodalityImagingforTumorVolumeDefinitioninRadiationOncology 79 Issam El Naqa 3.1 Introduction 79 3.2 Single versus Multimodality Image Segmentation 80 3.3 Methods for Multimodality Image Segmentation 82 3.3.1 Multiple-image thresholding 83 3.3.2 Clustering algorithms 83 3.3.2.1 Fuzzy C-means algorithm 83 3.3.2.2 Extending the fuzzy C-means algorithm to multiple images 84 3.3.2.3 K-means clustering algorithm 85 3.3.3 Active contour algorithms 85 3.3.3.1 “Active-contour-without-edge” algorithm 87 3.3.3.2 Extension to multiple images 88 3.4 Examples of Multimodality Tumor Volume Definition 89 3.4.1 PET/CT target definition in radiotherapy 89 3.4.2 PET/CT segmentation of cervix cancer example 90 3.4.3 MR/CT segmentation of prostate cancer example 91 3.4.4 Coronary artery plaque MR image analysis 92 3.5 Issues, Problems, and Future Directions 92 3.5.1 Image understanding 93 3.5.2 Deformable image registration 94 3.6 Conclusions 94 References 94 4 Nonlinear Unsharp Masking for Enhancing Suspicious Regions in Mammograms 99 Yicong Zhou, C. L. Philip Chen, Sos S. Agaian, and Karen Panetta 4.1 Introduction 99 4.2 Background 102 4.2.1 Traditional unsharp masking 102 4.2.2 The RUM algorithm 103 4.2.3 The ANCE algorithm 104 TableofContents vii 4.2.4 The CLAHE algorithm 105 4.2.5 The DICE algorithm 105 4.2.6 The PLIP operations 105 4.3 Nonlinear Unsharp Masking 106 4.3.1 The new NLUM scheme 106 4.3.2 Discussion 108 4.4 New Enhancement Measure 109 4.4.1 Discussion 109 4.4.2 New enhancement measure: SDME 111 4.5 Simulation Results and Evaluations 111 4.5.1 Comparison of enhancement measures 111 4.5.2 Parameter optimization 113 4.5.3 Enhancement analysis 115 4.5.4 HVS-based analysis and visualization 115 4.5.5 Comparison of enhancement performance 116 4.5.6 ROC evaluation 119 4.6 Conclusion 120 References 121 5 Skin Lesion Extraction Based on Distance Histogram and Color Information 131 Jinshan Tang and Yanliang Gu 5.1 Introduction 131 5.2 Color-Based Skin Lesion Segmentation 133 5.2.1 Noise reduction 133 5.2.2 Adaptive Color Model Building 134 5.2.2.1 Color spaces 134 5.2.2.2 Adaptive color model building 135 5.2.3 Distance-histogram-based lesion extraction 136 5.2.3.1 Skin color detection 136 5.2.3.2 Adaptive thresholding 137 5.2.3.3 Morphological processing 138 5.3 Experimental Results 139 5.3.1 Noise reduction on synthetic images 139 5.3.2 Noise reduction on skin lesion images 140 5.3.3 Experimental results on skin lesion segmentation 141 5.3.4 Speeding up using a GPU 144 5.4 Conclusion 146 References 146 6 Geometric Incremental Support Vector Machine for Object Detection from Capsule Endoscopy Videos 149 Xiaohui Yuan, Mohamed Abouelenien, Balathasan Giritharan, Jianguo Liu, and Xiaojing Yuan 6.1 Introduction 149 vii viii TableofContents 6.2 Related Work 150 6.2.1 Related work on CE video analysis for automatic object detection 150 6.2.2 Related work on incremental learning using SVMs 151 6.3 Geometric Incremental Support Vector Machines 153 6.3.1 Geometric support vector machines 153 6.3.2 Geometric incremental support vector machine (GISVM) 154 6.4 Experimental Results and Discussion 158 6.4.1 Synthetic and benchmark data preparation 158 6.4.2 Parameter selection 159 6.4.3 Efficiency analysis 161 6.4.4 Accuracy analysis 163 6.4.5 Experiments with CE videos 164 6.5 Conclusion 166 References 167 7 Automated Melanoma Screening and Early Detection 173 Xiaojing Yuan, Ning Situ, Xiaohui Yuan, and George Zouridakis 7.1 Overview of Automated Melanoma Screening and Early Detection Systems 174 7.1.1 Optical imaging modalities for pigmented skin lesion image acquisition 174 7.1.2 Key functions of existing skin lesion classification systems 176 7.1.3 Review of representative skin cancer detection systems 178 7.1.4 Overview of skin lesions and commonly used criteria by clinicians 179 7.1.5 Pigmented skin lesion datasets 182 7.2 AutoScan: Automated Melanoma Screening and Early Detection 183 7.2.1 AutoScan preprocessing 184 7.2.2 AutoScan: region-of-interest identification 184 7.3 Mapping Computer-Generated Features to High-Level Concepts Used by Dermatologists 187 7.3.1 Computer-generated low-level features 188 7.3.2 Mappinghigh-leveldermoscopicconceptswithmultiple-instance learning 189 7.3.2.1 Diverse density function and evidence confidence function 190 7.3.2.2 Boosting enhanced-instance prototype selection 191 7.3.3 Transforming a lesion image into a descriptor vector 192 7.3.4 Experiment setups and results 192 7.4 Integrating Feature Selection with Feature-Model Learning 194 7.4.1 Feature selection and combination 195 7.4.2 Multiple auxiliary kernel learning (MAKL): learning from heterogeneous feature spaces 196 7.4.3 Experiment setup and results 198 TableofContents ix 7.5 Conclusions and Future Directions 201 References 201 8 AComplexWavelet-BasedFeatureExtractionSystemforMicrocalcification DetectioninDigitalMammograms 211 Ping Zhang and Kwabena Agyepong 8.1 Introduction 211 8.2 System Design 213 8.3 Hybrid Feature Extraction 214 8.3.1 Surrounding region dependence-based method 214 8.3.2 Wavelet transform 216 8.3.3 Complex wavelet transform for feature extraction 217 8.3.4 2D-CWT multifractal feature 219 8.4 Classifier Design 220 8.5 Experiment Results 220 8.6 Conclusion 224 References 224 9 Computer-AidedProstateCancerDiagnosis:Principles,RecentAdvances, andFutureProspective 229 Sos S. Agaian, Clara Mosquera-Lopez, Alejandro Velez-Hoyos, and Ian Thompson 9.1 Introduction 229 9.2 Clinical Approach for Prostate Cancer Detection and Grading 234 9.2.1 Core needle biopsy 234 9.2.2 Digital pathology imaging 235 9.3 State-of-the-Art in Histopathology-Image-Based, Computer-Aided Prostate Cancer Diagnosis 236 9.3.1 Image preprocessing 238 9.3.1.1 Color normalization 238 9.3.1.2 Histopathology image segmentation 239 9.3.2 Feature extraction 243 9.3.3 Classification 247 9.3.4 System accuracy assessment 248 9.3.4.1 Performance indicators 249 9.4 Conclusions, Future Directions, and Potential New Strategies 253 References 256 10 AnalysisofBreastMassesinMammogramsUsingtheFractalDimension andShapeFactors 269 Grazia Raguso, Antonietta Ancona, Loredana Chieppa, Samuela L’Abbate, Maria Luisa Pepe, Fabio Mangieri, Shantanu Banik, and Rangaraj M. Rangayyan 10.1 Introduction 270 10.2 Methods 273 10.2.1 Fractal analysis 273 x TableofContents 10.2.2 Shape factors 276 10.2.3 Feature analysis, selection, and classification 279 10.3 Datasets of Contours of Breast Masses 280 10.4 Results and Discussion 282 10.5 Conclusion 286 References 286 11 Another Step towards Successful Tomographic Imaging in Cancer: Solving the Problem of Image Reconstruction 295 Artyom M. Grigoryan 11.1 Introduction 295 11.1.1 CT images and lung cancer 296 11.1.2 CT images and breast cancer 296 11.1.3 Breast cancer with CT, mammography, and MRI 297 11.1.4 Algorithms in CT 297 11.2 Model of the Image 298 11.2.1 Line integrals and ray sums 299 11.2.2 Two types of parallel rays 300 11.3 The Image and the Set of Splitting-Signals 301 11.4 Geometry of the Projections on the Lattice 305 11.4.1 Main equations for geometrical rays when N is prime 315 11.4.2 Simulation results for modeled images 316 11.5 Geometry for the Lattice N (cid:1) N when N is a Power of Two 318 11.5.1 Algorithm of image reconstruction 326 11.5.2 Convolution equations 327 11.5.3 Preliminary results 329 11.6 Conclusion 333 References 334 Index 339