Lu / Zheng / Carneiro Deep Learning and Convolutional Neural Networks for Medical Image Computing

Precision Medicine, High Performance and Large-Scale Datasets
1. Auflage 2017
ISBN: 978-3-319-42999-1
Verlag: Springer Nature Switzerland
Format: PDF
Kopierschutz: 1 - PDF Watermark

E-Book, Englisch, 327 Seiten

Reihe: Advances in Computer Vision and Pattern Recognition

This book presents a detailed review of the state of the art in deep learning approaches for semantic object detection and segmentation in medical image computing, and large-scale radiology database mining. A particular focus is placed on the application of convolutional neural networks, with the theory supported by practical examples. Features: highlights how the use of deep neural networks can address new questions and protocols, as well as improve upon existing challenges in medical image computing; discusses the insightful research experience of Dr. Ronald M. Summers; presents a comprehensive review of the latest research and literature; describes a range of different methods that make use of deep learning for object or landmark detection tasks in 2D and 3D medical imaging; examines a varied selection of techniques for semantic segmentation using deep learning principles in medical imaging; introduces a novel approach to interleaved text and image deep mining on a large-scale radiology image database.


Dr. Le Lu is a Staff Scientist in the Radiology and Imaging Sciences Department of the National Institutes of Health Clinical Center, Bethesda, MD, USA.
Dr. Yefeng Zheng is a Senior Staff Scientist at Siemens Healthcare Technology Center, Princeton, NJ, USA.
Dr. Gustavo Carneiro is an Associate Professor in the School of Computer Science at The University of Adelaide, Australia.
Dr. Lin Yang is an Associate Professor in the Department of Biomedical Engineering at the University of Florida, Gainesville, FL, USA.

Autoren/Hrsg.

Lu, Le

Zheng, Yefeng

Carneiro, Gustavo

Yang, Lin

Inhaltsverzeichnis

1;Preface;6
1.1;Overview and Goals;7
1.2;Organization and Features;8
1.3;Target Audience;9
2;Acknowledgements;10
3;Contents;11
4;Part I Review;14
5;1 Deep Learning and Computer-Aided Diagnosis for Medical Image Processing: A Personal Perspective;15
5.1;References;20
6;2 Review of Deep Learning Methods in Mammography, Cardiovascular, and Microscopy Image Analysis;23
6.1;2.1 Introduction on Deep Learning Methods in Mammography;23
6.2;2.2 Deep Learning Methods in Mammography;24
6.3;2.3 Summary on Deep Learning Methods in Mammography;26
6.4;2.4 Introduction on Deep Learning for Cardiological Image Analysis;26
6.5;2.5 Deep Learning-Based Methods for Heart Segmentation;28
6.6;2.6 Deep Learning-Based Methods for Vessel Segmentation;29
6.7;2.7 Introduction to Microscopy Image Analysis;31
6.8;2.8 Deep Learning Methods;33
6.9;2.9 Microscopy Image Analysis Applications;34
6.10;2.10 Discussions and Conclusion on Deep Learning for Microscopy Image Analysis;34
6.11;References;38
7;Part II Detection and Localization;45
8;3 Efficient False Positive Reduction in Computer-Aided Detection Using Convolutional Neural Networks and Random View Aggregation;46
8.1;3.1 Introduction;47
8.2;3.2 Related Work;47
8.2.1;3.2.1 Cascaded Classifiers in CADe;48
8.3;3.3 Methods;48
8.3.1;3.3.1 Convolutional Neural Networks;48
8.3.2;3.3.2 A 2D or 2.5D Approach for Applying ConvNets to CADe;50
8.3.3;3.3.3 Random View Aggregation;52
8.3.4;3.3.4 Candidate Generation;52
8.4;3.4 Results;53
8.4.1;3.4.1 Computer-Aided Detection Data Sets;53
8.5;3.5 Discussion and Conclusions;54
8.6;References;56
9;4 Robust Landmark Detection in Volumetric Data with Efficient 3D Deep Learning;60
9.1;4.1 Introduction;60
9.2;4.2 Training Shallow Network with Separable Filters;63
9.3;4.3 Training Sparse Deep Network;66
9.4;4.4 Robust Detection by Combining Multiple Features;67
9.5;4.5 Experiments;68
9.6;4.6 Conclusions;70
9.7;References;71
10;5 A Novel Cell Detection Method Using Deep Convolutional Neural Network and Maximum-Weight Independent Set;73
10.1;5.1 Introduction;73
10.2;5.2 Methodology;75
10.2.1;5.2.1 Cell Detection Using MWIS;75
10.2.2;5.2.2 Deep Convolutional Neural Network;76
10.3;5.3 Experiments;78
10.4;5.4 Conclusion;81
10.5;References;81
11;6 Deep Learning for Histopathological Image Analysis: Towards Computerized Diagnosis on Cancers;83
11.1;6.1 Introduction;84
11.2;6.2 Previous Works;85
11.2.1;6.2.1 Previous Works on Deep Learning for Histological Image Analysis;86
11.2.2;6.2.2 Previous Works on Nuclear Atypia Scoring;87
11.2.3;6.2.3 Previous Works on Epithelial and Stromal Segmentation;88
11.3;6.3 Deep Learning for Nuclear Atypia Scoring;88
11.3.1;6.3.1 CN Model for Nuclear Atypia Scoring;90
11.3.2;6.3.2 Integration MR-CN with Combination Voting Strategies for NAS;91
11.4;6.4 Deep Learning for Epithelial and Stromal Tissues Segmentation;94
11.4.1;6.4.1 The Deep Convolutional Neural Networks;94
11.4.2;6.4.2 Generating Training and Testing Samples;94
11.4.3;6.4.3 The Trained CN for the Discrimination of EP and ST Regions;95
11.5;6.5 Experimental Setup;96
11.5.1;6.5.1 Data Set;97
11.5.2;6.5.2 Comparison Strategies;98
11.5.3;6.5.3 Computational and Implemental Consideration;99
11.6;6.6 Results and Discussion;99
11.6.1;6.6.1 Qualitative Results;99
11.6.2;6.6.2 Quantitative Results;100
11.7;6.7 Concluding Remarks;102
11.8;References;102
12;7 Interstitial Lung Diseases via Deep Convolutional Neural Networks: Segmentation Label Propagation, Unordered Pooling and Cross-Dataset Learning;106
12.1;7.1 Introduction;106
12.2;7.2 Methods;109
12.2.1;7.2.1 Segmentation Label Propagation;110
12.2.2;7.2.2 Multi-label ILD Regression;112
12.3;7.3 Experiments and Discussion;114
12.3.1;7.3.1 Segmentation Label Propagation;114
12.3.2;7.3.2 Multi-label ILD Regression;116
12.4;7.4 Conclusion;118
12.5;References;119
13;8 Three Aspects on Using Convolutional Neural Networks for Computer-Aided Detection in Medical Imaging;121
13.1;8.1 Introduction;122
13.2;8.2 Datasets and Related Work;124
13.3;8.3 Methods;127
13.3.1;8.3.1 Convolutional Neural Network Architectures;127
13.3.2;8.3.2 ImageNet: Large-Scale Annotated Natural Image Dataset;129
13.3.3;8.3.3 Training Protocols and Transfer Learning;129
13.4;8.4 Experiments and Discussions;131
13.4.1;8.4.1 Thoracoabdominal Lymph Node Detection;131
13.4.2;8.4.2 Interstitial Lung Disease Classification;133
13.4.3;8.4.3 Evaluation of Five CNN Models Using ILD Classification;137
13.4.4;8.4.4 Analysis via CNN Learning Visualization;138
13.4.5;8.4.5 Findings and Observations;139
13.5;8.5 Conclusion;140
13.6;References;141
14;9 Cell Detection with Deep Learning Accelerated by Sparse Kernel;145
14.1;9.1 Introduction;145
14.1.1;9.1.1 Related Work;146
14.1.2;9.1.2 Challenges;152
14.2;9.2 Pixel-Wise Cell Detector;153
14.2.1;9.2.1 Overview;153
14.2.2;9.2.2 Deep Convolutional Neural Network;154
14.2.3;9.2.3 Implementation;155
14.3;9.3 Sparse Kernel Acceleration of the Pixel-Wise Cell Detector;156
14.3.1;9.3.1 Training the Detector;156
14.3.2;9.3.2 Deep Convolution Neural Network Architecture;156
14.3.3;9.3.3 Acceleration of Forward Detection;157
14.4;9.4 Experiments;158
14.4.1;9.4.1 Materials and Experiment Setup;158
14.4.2;9.4.2 Results;160
14.5;9.5 Discussion;160
14.6;9.6 Conclusion;162
14.7;References;163
15;10 Fully Convolutional Networks in Medical Imaging: Applications to Image Enhancement and Recognition;166
15.1;10.1 Introduction;166
15.2;10.2 Image Super-Resolution;168
15.2.1;10.2.1 Motivation;168
15.2.2;10.2.2 Methodology;169
15.2.3;10.2.3 Results;172
15.2.4;10.2.4 Discussion and Conclusion;175
15.3;10.3 Scan Plane Detection;175
15.3.1;10.3.1 Motivation;175
15.3.2;10.3.2 Materials and Methods;177
15.3.3;10.3.3 Experiments and Results;181
15.3.4;10.3.4 Discussion and Conclusion;182
15.4;10.4 Discussion and Conclusion;184
15.5;References;185
16;11 On the Necessity of Fine-Tuned Convolutional Neural Networks for Medical Imaging;187
16.1;11.1 Introduction;188
16.2;11.2 Related Works;188
16.3;11.3 Contributions;190
16.4;11.4 Applications and Results;190
16.4.1;11.4.1 Polyp Detection;192
16.4.2;11.4.2 Pulmonary Embolism Detection;193
16.4.3;11.4.3 Colonoscopy Frame Classification;194
16.4.4;11.4.4 Intima-Media Boundary Segmentation;195
16.5;11.5 Discussion;196
16.6;11.6 Conclusion;197
16.7;References;198
17;Part III Segmentation;200
18;12 Fully Automated Segmentation Using Distance Regularised Level Set and Deep-Structured Learning and Inference;201
18.1;12.1 Introduction;202
18.2;12.2 Literature Review;203
18.3;12.3 Methodology;205
18.3.1;12.3.1 Left Ventricle Segmentation;208
18.3.2;12.3.2 Endocardium Segmentation;209
18.3.3;12.3.3 Epicardium Segmentation;213
18.3.4;12.3.4 Lung Segmentation;214
18.4;12.4 Experiments;214
18.4.1;12.4.1 Data Sets and Evaluation Measures;214
18.4.2;12.4.2 Experimental Setup;215
18.4.3;12.4.3 Results of Each Stage of the Proposed Methodology;217
18.4.4;12.4.4 Comparison with the State of the Art;217
18.5;12.5 Discussion and Conclusions;219
18.6;References;227
19;13 Combining Deep Learning and Structured Prediction for Segmenting Masses in Mammograms;229
19.1;13.1 Introduction;229
19.2;13.2 Literature Review;231
19.3;13.3 Methodology;232
19.3.1;13.3.1 Conditional Random Field (CRF);233
19.3.2;13.3.2 Structured Support Vector Machine (SSVM);235
19.3.3;13.3.3 Potential Functions;236
19.4;13.4 Experiments;239
19.4.1;13.4.1 Materials and Methods;239
19.4.2;13.4.2 Results;240
19.5;13.5 Discussion and Conclusions;242
19.6;References;243
20;14 Deep Learning Based Automatic Segmentation of Pathological Kidney in CT: Local Versus Global Image Context;245
20.1;14.1 Introduction;245
20.2;14.2 Training Data Synthesis;248
20.3;14.3 Abdomen Localization Using Deep Learning;250
20.4;14.4 Kidney Localization Using Deep Learning;253
20.5;14.5 Kidney Segmentation Based on MSL;254
20.6;14.6 Experiments;255
20.7;14.7 Conclusions;258
20.8;References;258
21;15 Robust Cell Detection and Segmentation in Histopathological Images Using Sparse Reconstruction and Stacked Denoising Autoencoders;260
21.1;15.1 Introduction;261
21.2;15.2 Methodology;263
21.2.1;15.2.1 Detection via Sparse Reconstruction with Trivial Templates;264
21.2.2;15.2.2 Cell Segmentation via Stacked Denoising Autoencoders;265
21.2.3;15.2.3 The Learned Filters;267
21.2.4;15.2.4 Training DAE with Discriminative Loss;268
21.3;15.3 Experimental Results;273
21.3.1;15.3.1 Computational Complexity;278
21.4;15.4 Conclusion;279
21.5;References;279
22;16 Automatic Pancreas Segmentation Using Coarse-to-Fine Superpixel Labeling;282
22.1;16.1 Introduction;283
22.2;16.2 Previous Literature;285
22.3;16.3 Methods;286
22.3.1;16.3.1 Boundary-Preserving Over-segmentation;286
22.3.2;16.3.2 Patch-Level Visual Feature Extraction and Classification: PRF;289
22.3.3;16.3.3 Patch-Level Labeling via Deep Convolutional Neural Network: PCNN;292
22.3.4;16.3.4 Superpixel-Level Feature Extraction, Cascaded Classification, and Pancreas Segmentation;293
22.4;16.4 Data and Experimental Results;295
22.4.1;16.4.1 Imaging Data;295
22.4.2;16.4.2 Experiments;296
22.5;16.5 Conclusion and Discussion;302
22.6;References;303
23;Part IV Big Dataset and Text-Image Deep Mining;306
24;17 Interleaved Text/Image Deep Mining on a Large-Scale Radiology Image Database;307
24.1;17.1 Introduction;307
24.1.1;17.1.1 Related Work;309
24.2;17.2 Data;310
24.3;17.3 Document Topic Learning with Latent Dirichlet Allocation;311
24.4;17.4 Image to Document Topic Mapping with Deep Convolutional Neural Networks;313
24.5;17.5 Generating Image-to-Text Description;315
24.5.1;17.5.1 Removing Word-Level Ambiguity with Word-to-Vector Modeling;316
24.5.2;17.5.2 Using Sentences to Words Based Image Representation;317
24.5.3;17.5.3 Bi-gram Deep CNN Regression;317
24.5.4;17.5.4 Word Prediction from Images as Retrieval;318
24.6;17.6 Conclusion and Discussion;320
24.7;References;321
25;Author Index;324
26;Subject Index;326