Tongue Image Analysis (eBook)

David Zhang graduated in Computer Science from Peking University. He received his MSc in 1982 and his PhD in 1985 in Computer Science from the Harbin Institute of Technology (HIT). From 1986 to 1988 he was a postdoctoral fellow at Tsinghua University and then an associate professor at the Academia Sinica, Beijing. In 1994 he received his second PhD in Electrical and Computer Engineering from the University of Waterloo, Ontario, Canada. Currently, he is a Chair Professor at the Hong Kong Polytechnic University, where he is the Founding Director of Biometrics Research Centre (UGC/CRC), which has been supported by the Hong Kong SAR Government since 1998. He also serves as Visiting Chair Professor at Tsinghua University and HIT, and Adjunct Professor at Shanghai Jiao Tong University, Peking University, the National University of Defense Technology and the University of Waterloo. So far, he has been published more than 10 books and 400 international journal papers. He was listed as a highly cited researcher in Engineering by Thomson Reuters in 2014 and in 2015, respectively. Professor Zhang is a Croucher senior research fellow, distinguished speaker of the IEEE computer society, and a Fellow of both the IEEE and the IAPR.Hongzhi Zhang received his Ph.D. degree in computer science and technology from Harbin institute of Technology (HIT), China, in 2007. He is an associate Professor at the School of Computer Science and Technology, HIT, where he has taught for over 15 years. He teaches biomedical image processing and has investigated computerized tongue diagnosis at the Research Center of Perception and Computing. He is a member of  the IEEE, the Chinese Association for Artificial Intelligence (CAAI), and the  China Society of Integrated Traditional Chinese and Western Medicine. His research interests include theoretic approaches to problems in biomedical imaging, biometric image analysis, computer vision, and signal processing. His research has been supported by grants from the National Natural Science Foundation of China. He is the author of more than 70 international journal and conference papers.  Bob Zhang received his Ph.D. degree in electrical and computer engineering from the University of Waterloo, Waterloo, Canada, in 2011. After graduating from Waterloo, he remained with the Center for Pattern Recognition and Machine Intelligence, and was later a postdoctoral researcher at the Department of Electrical and Computer Engineering at Carnegie Mellon University, Pittsburgh, USA. He is currently an assistant professor in the Department of Computer and Information Science, University of Macau, Taipa, Macau. His research interests focus on medical biometrics, biometrics security, pattern recognition, and image processing. Dr. Zhang is a Technical Committee Member of the IEEE Systems, Man, and Cybernetics Society, an Associate Editor for the International Journal of Image and Graphics, as well as an editorial board member for the International Journal of INFORMATION.

Preface 5
Contents 9
Background 16
1 Introduction to Tongue Image Analysis 17
Abstract 17
1.1 Tongue Inspection for Medical Applications 17
1.2 Computerized Tongue Diagnosis System 20
1.3 Research Review on Tongue Image Analysis 21
1.3.1 Tongue Image Acquisition 21
1.3.2 Tongue Image Preprocessing 22
1.3.2.1 Color Correction 22
1.3.2.2 Image Segmentation 23
1.3.3 Qualitative Feature Extraction 24
1.3.4 Diagnostic Classification 25
1.4 Issues and Challenges 25
1.4.1 Inconsistent Image Acquisition 26
1.4.2 Inaccurate Color Correction 27
1.4.3 Subjective Tongue Color Extraction and Classification 28
References 28
2 Tongue Images Acquisition System Design 33
Abstract 33
2.1 Introduction 33
2.2 System Framework and Requirement Analysis 36
2.2.1 System Framework 37
2.2.2 Requirement Analysis 38
2.3 Optimal System Design 41
2.3.1 Illuminant 41
2.3.2 Lighting Condition 42
2.3.3 Imaging Camera 44
2.3.4 Color Correction 46
2.3.5 System Implementation and Calibration 47
2.3.5.1 Camera Lens Aperture 48
2.3.5.2 Camera Color Balance 49
2.3.5.3 Color Correction Model 49
2.4 Performance Analysis 49
2.4.1 Illumination Uniformity 50
2.4.2 System Consistency 51
2.4.2.1 Consecutive Consistency 52
2.4.2.2 Intra-run Consistency 52
2.4.2.3 Inter-run Consistency 53
2.4.2.4 Between-Device Consistency 54
2.4.3 Accuracy 55
2.4.4 Typical Tongue Images 55
2.5 Summary 56
References 57
Tongue Image Segmentation and Shape Classification 59
3 Tongue Image Segmentation by Bi-elliptical Deformable Contour 60
Abstract 60
3.1 Introduction 60
3.2 Bi-elliptical Deformable Template for the Tongue 63
3.2.1 Definitions and Notations 63
3.2.2 The Tongue Template 64
3.2.3 Energy Function for the Tongue Template 65
3.3 Combined Model for Tongue Segmentation 68
3.3.1 Two Kinds of Template Forces 69
3.3.1.1 Linear Template Force (LTF) 70
3.3.1.2 Elliptical Template Force (ETF) 71
3.3.2 Bi-elliptical Deformable Contours 72
3.3.2.1 Tongue Segmentation Algorithm 73
3.4 Experiment Results and Analysis 75
3.5 Summary 82
References 82
4 A Snake-Based Approach to Automated Tongue Image Segmentation 84
Abstract 84
4.1 Introduction 84
4.2 Automated Segmentation Algorithm for Tongue Images 86
4.2.1 Polar Edge Detection of Tongue Image 86
4.2.2 Filtering and Binarization of the Edge Image 88
4.2.3 Initialization and ACM 89
4.2.4 Summary of the Automated Tongue Segmentation Method 91
4.3 Experiments and Discussion 93
4.3.1 Evaluation on the Edge Filtering Algorithm 93
4.3.2 Qualitative Evaluation 93
4.3.3 Quantitative Evaluation 95
4.3.3.1 Boundary Error Metrics 96
4.3.3.2 Area Error Metrics 98
4.4 Summary 100
References 100
5 Tongue Segmentation in Hyperspectral Images 102
Abstract 102
5.1 Introduction 102
5.2 Setup of the Hyperspectral Device 104
5.3 Segmentation Framework 105
5.3.1 Hyperspectral Image Calibration 106
5.3.2 Segmentation 107
5.4 Experiments and Comparisons 109
5.4.1 Criteria of Evaluation 111
5.4.2 Comparison with the BEDC 112
5.5 Summary 114
References 114
6 Tongue Segmentation by Gradient Vector Flow and Region Merging 116
Abstract 116
6.1 Introduction 116
6.2 Initial Segmentation 117
6.3 Extraction of Tongue Area 119
6.3.1 Similarity Metric 119
6.3.2 The Extraction of the Tongue Body by Using the MRSM Algorithm 120
6.4 Experimental Results and Discussions 121
6.4.1 Experimental Results 121
6.4.2 Qualitative Evaluation 122
6.4.3 Quantitative Evaluation 123
6.4.4 Running Time of the Proposed Method 124
6.4.5 Limitations of the Proposed Method 125
6.5 Summary 125
References 126
7 Tongue Segmentation by Fusing Region-Based and Edge-Based Approaches 127
Abstract 127
7.1 Introduction 127
7.2 Extraction of the ROI to Enhance Robustness 129
7.3 Combining Region-Based and Edge-Based Approaches 132
7.3.1 Region-Based Approach: Improved MSRM 133
7.3.2 Optimal Edge-Based Approach: Fast Marching 135
7.3.3 The Fusion Approach as a Solution 137
7.4 Experiments and Comparisons 139
7.4.1 Qualitative Evaluation 139
7.4.2 Quantitative Evaluation 141
7.5 Summary 142
References 142
8 Tongue Shape Classification by Geometric Features 144
Abstract 144
8.1 Introduction 144
8.2 Shape Correction 145
8.2.1 Automatic Contour Extraction 146
8.2.2 The Length Criterion 146
8.2.3 The Area Criterion 147
8.2.4 The Angle Criterion 148
8.2.5 Correction by Combination 149
8.3 Feature Extraction 150
8.3.1 The Length-Based Feature 151
8.3.1.1 The Length–Width Ratio 151
8.3.1.2 The off-Center Ratio 151
8.3.1.3 The Radial Line Ratio 152
8.3.2 The Area-Based Feature 152
8.3.2.1 The Total Area Ratio 152
8.3.2.2 The Triangle Area Ratio 153
8.3.2.3 The Top–Bottom Area Ratio 154
8.3.3 The Angle-Based Feature 154
8.4 Shape Classification 155
8.4.1 Modeling the Classification as a Hierarchy 155
8.4.2 Calculating Relative Weights 157
8.4.3 Calculating the Global Weights 158
8.4.4 Fuzzy Shape Classification 158
8.5 Experimental Results and Performance Analysis 159
8.5.1 Accuracy of Shape Correction 159
8.5.2 Accuracy of Shape Classification 160
8.6 Summary 163
References 163
Tongue Color Correction and Classification 165
9 Color Correction Scheme for Tongue Images 166
Abstract 166
9.1 Introduction 166
9.2 Color Space for Tongue Analysis 168
9.3 Color Correction Algorithms 170
9.3.1 Definitions of Algorithms 171
9.3.2 Evaluation of the Correction Algorithms 172
9.3.3 Experiments and Results 173
9.3.3.1 Imaging Device 173
9.3.3.2 Dataset 173
9.3.3.3 Parameters Setting 174
9.3.3.4 Experimental Results on the Colorchecker 175
9.3.3.5 Experimental Results on Tongue Images 176
9.3.3.6 Discussion 178
9.4 Experimental Results and Performance Analysis 178
9.4.1 Color Correction by Different Cameras 179
9.4.2 Color Correction Under Different Lighting Conditions 180
9.4.3 Performance Analysis 182
9.4.4 Correction on Real Tongue Images 183
9.5 Summary 185
References 186
10 Tongue Colorchecker for Precise Correction 188
Abstract 188
10.1 Introduction 188
10.2 Tongue Color Space 190
10.3 Determination of the Number of Colors 192
10.3.1 Setting for Number Deciding Experiment 193
10.3.1.1 Training and Testing Dataset 193
10.3.1.2 Experimental Implementation 194
10.3.1.3 Configuration for Stability Test 195
10.3.2 Results of Number Determination 196
10.3.2.1 Obtained Minimum Sufficient Number 196
10.3.2.2 Results of the Stability Test 197
10.4 Optimal Colors Selection 199
10.4.1 Objective Function 199
10.4.2 Selection Algorithms 201
10.4.2.1 The Greedy Method 201
10.4.2.2 The Clustering-Based Method 202
10.4.2.3 Selection the Color Space 203
10.5 Experimental Results and Performance Analysis 204
10.5.1 Experimental Configuration 204
10.5.1.1 Training and Testing Dataset 204
10.5.1.2 Flowchart of the Experiment 204
10.5.2 Parameter Optimization 205
10.5.2.1 Color Space 205
10.5.2.2 Selection Method 207
10.5.2.3 Diversity Measurement 208
10.5.2.4 Distance Measure 209
10.5.2.5 Performance Comparison 211
10.6 Summary 213
References 213
11 Tongue Color Analysis for Medical Application 215
Abstract 215
11.1 Introduction 215
11.2 Tongue Image Acquisition Device and Dataset 217
11.3 Tongue Color Gamut and Color Features Extraction 218
11.3.1 Tongue Color Gamut 218
11.3.2 Tongue Color Features 220
11.4 Results and Discussion 223
11.4.1 Healthy Versus Disease Classification 223
11.4.2 Typical Disease Analysis 224
11.5 Summary 230
References 231
12 Statistical Analysis of Tongue Color and Its Applications in Diagnosis 232
Abstract 232
12.1 Introduction 232
12.2 Tongue Image Acquisition and Database 234
12.2.1 Tongue Image Acquisition Device 234
12.2.2 Color Correction of Tongue Images 235
12.2.3 Tongue Image Database 237
12.3 Tongue Color Distribution Analysis 238
12.3.1 Tongue Color Gamut: Generation and Modeling 238
12.3.2 Tongue Color Centers 246
12.3.3 Distribution of Typical Image Features 249
12.4 Color Feature Extraction 251
12.4.1 Tongue Color Feature Vector 252
12.4.2 Typical Samples of Tongue Color Representation 252
12.5 Summary 255
References 255
13 Hyperspectral Tongue Image Classification 258
Abstract 258
13.1 Introduction 258
13.2 Hyperpectral Images for Tongue Diagnosis 260
13.3 The Classifier Applied to Hyperspectral Tongue Images 261
13.3.1 Linear SVM: Linearly Separable 261
13.3.2 Linear SVM: Linearly Non-separable 262
13.3.3 Non-linear SVM 263
13.4 Experimental Results and Performance Analysis 264
13.4.1 Comparing Linear and Non-linear SVM, RBFNN, and K-NN Classifiers 264
13.4.2 Evaluating the Diagnostic Performance of SVM 265
13.5 Summary 267
References 268
Tongue Image Analysis and Diagnosis 269
14 Computerized Tongue Diagnosis Based on Bayesian Networks 270
Abstract 270
14.1 Introduction 270
14.2 Tongue Diagnosis Using Bayesian Networks 271
14.3 Quantitative Pathological Features Extraction 274
14.3.1 Quantitative Color Features 274
14.3.2 Quantitative Texture Features 275
14.4 Experimental Results 277
14.4.1 Several Issues 278
14.4.2 Bayesian Network Classifier Based on Textural Features 279
14.4.3 Bayesian Network Classifier Based on Chromatic Features 280
14.4.4 Bayesian Network Classifier Based on Combined Features 281
14.5 Summary 284
References 284
15 Tongue Image Analysis for Appendicitis Diagnosis 286
Abstract 286
15.1 Introduction 286
15.2 Chromatic and Textural Features for Tongue Diagnosis 287
15.2.1 The Image of the Tongue of a Patient with Appendicitis 287
15.2.2 Quantitative Features of the Color of the Tongue 288
15.2.3 Quantitative Features of the Texture of the Tongue 288
15.3 Identification of Filiform Papillae 289
15.3.1 Typical Figures and Statistics of Filiform Papillae 289
15.3.2 Filter for Filiform Papillae 291
15.4 Experimental Results and Analysis 292
15.4.1 Evaluation Basis for Diagnosis 293
15.4.2 Performance of Metrics for Color 293
15.4.3 Performance of Textural Metrics 295
15.4.4 Performance of the FPF 296
15.5 Summary 298
References 298
16 Diagnosis Using Quantitative Tongue Feature Classification 299
Abstract 299
16.1 Introduction 299
16.2 Tongue Image Samples 300
16.3 Quantitative Chromatic and Textural Measurements 300
16.4 Feature Selection 302
16.5 Results and Analysis 302
16.6 Summary 303
References 305
17 Detecting Diabetes Mellitus and Nonproliferative Diabetic Retinopathy Using CTD 306
Abstract 306
17.1 Introduction 306
17.2 Capture Device and Tongue Image Preprocessing 308
17.3 Tongue Color Features 309
17.3.1 Tongue Color Gamut 309
17.3.2 Color Feature Extraction 310
17.4 Tongue Texture Features 314
17.5 Tongue Geometric Features 316
17.6 Numerical Results and Discussion 320
17.6.1 Healthy Versus DM Classification 320
17.6.2 NPDR Versus DM-Sans NPDR Classification 324
17.7 Summary 326
References 327
Book Recapitulation 329
18 Book Review and Future Work 330
Abstract 330
18.1 Book Recapitulation 330
18.2 Future Work 331
Index 333