Menu Expand
Medical Biometrics: Computerized Tcm Data Analysis

Medical Biometrics: Computerized Tcm Data Analysis

Zhang David | Zuo Wangmeng

(2016)

Additional Information

Book Details

Abstract

The introduction of traditional Chinese medicine (TCM) through modern information technology will not only achieve the objective progress of the heritage of thousands of years of TCM, but also deliver novel discoveries for modern medicines.This book is an advanced monograph based on a decade's worth of research work by the authors. After a brief introduction on the four diagnosis approaches in TCM, this book delves into the three main TCM data analysis techniques: computerized tongue, pulse and odor analysis.Both graduate students and researchers in computerized TCM data analysis will benefit from the book as it will provide a comprehensive understanding of the state-of-the-art analysis methods, image / signal acquisition devices, and the related feature extraction and classification methods.

Table of Contents

Section Title Page Action Price
Contents ix
Preface v
PART I: DIAGNOSIS METHODS IN TRADITIONAL CHINESE MEDICINE 1
Chapter 1 Introduction 3
1.1 Diagnosis Methods in Traditional Chinese Medicine 3
1.1.1 Tongue Diagnosis 3
1.1.2 Pulse Diagnosis 5
1.1.3 Breath Odor Diagnosis 6
1.2 Computerized TCM Diagnosis 7
1.2.1 Computerized Tongue Diagnosis 7
1.2.1.1 Tongue Image Acquisition 8
1.2.1.2 Tongue Image Preprocessing 8
1.2.1.3 Qualitative Feature Extraction 10
1.2.1.4 Diagnostic Classification 11
1.2.2 Computerized Pulse Diagnosis 11
1.2.2.1 Pulse Signal Acquisition 12
1.2.2.2 Baseline Wander Correction 12
1.2.2.3 Pulse Feature Extraction 12
1.2.2.4 Pulse Diagnosis 13
1.2.3 Computerized Breath Odor Diagnosis 14
1.2.3.1 Breath Odor Acquisition and Preprocessing 14
1.2.3.2 Feature Extraction 16
1.2.3.3 Classification and Diagnosis 16
1.3 Summary 17
References\r 17
PART II: COMPUTERIZED TONGUE IMAGE ANALYSIS 27
Chapter 2 Tongue Image Acquisition and Preprocessing 29
2.1 Tongue Image Acquisition 29
2.1.1 Requirement Analysis 31
2.1.2 System Design and Implementation 33
2.1.2.1 Illuminant 35
2.1.2.2 Lighting Condition 37
2.1.2.3 Camera 39
2.1.2.4 System Implementation 41
2.1.3 Performance Analysis 43
2.1.3.1 Illumination Uniformity 43
2.1.3.2 System Consistency 44
2.1.3.3 Accuracy 48
2.1.3.4 Typical Tongue Images 48
2.2 Color Correction 49
2.2.1 Color Correction Algorithms 51
2.2.2 Evaluation of Correction Algorithms 53
2.2.3 Discussion 61
2.2.3.1 Color Correction for Images Acquired by Different Cameras 62
2.2.3.2 Color Correction by Different Lighting Conditions 63
2.2.3.3 Performance Analysis 65
2.2.3.4 Correction on Real Tongue Images 65
2.3 Summary 67
References\r 68
Chapter 3 Automated Tongue Segmentation 73
3.1 Bi-Elliptical Deformable Contour 73
3.1.1 Bi-Elliptical Deformable Template for the Tongue 74
3.1.1.1 Definitions and Notations 74
3.1.1.2 The Tongue Template 75
3.1.1.3 Energy Function for the Tongue Template 76
3.1.2 Combined Model for Tongue Segmentation 78
3.1.2.1 Two Kinds of Template Forces 79
3.1.2.2 Bi-Elliptical Deformable Contours 82
3.1.2.3 Tongue Segmentation Algorithm 83
3.1.3 Results and Analysis 84
3.2 Snake with Polar Edge Detector 91
3.2.1 The Segmentation Algorithm 91
3.2.1.1 Polar Edge Detection of Tongue Image 92
3.2.1.2 Filtering and Binarization of Edge Image 94
3.2.1.3 Initialization and Active Contour Model 96
3.2.1.4 Summary of the Automated Tongue Segmentation Method 97
3.2.2 Experimental Results 99
3.2.2.1 Evaluation on the Edge-Filtering Algorithm 99
3.2.2.2 Qualitative Evaluation 99
3.2.2.3 Quantitative Evaluation 99
3.3 Gabor Magnitude-based Edge Detection and Fast Marching 104
3.3.1 2D Gabor Magnitude-based Edge Detection 105
3.3.1.1 2D Gabor Magnitude - based Detector 105
3.3.1.2 Edge Thresholding 108
3.3.2 Contour Detection Using Fast Marching and Active Contour Model 109
3.3.2.1 Selection of Stable Segments 109
3.3.2.2 Contour Initialization Using Fast Marching 109
3.3.2.3 Gradient Vector Flow Snake 111
3.3.3 Experimental Results 111
3.4 Summary 114
References\r 114
Chapter 4 Tongue Image Feature Analysis 117
4.1 Color Feature Analysis 117
4.1.1 Exploratory Tongue Color Analysis 118
4.1.1.1 Color Space 118
4.1.1.2 Tongue Color Analysis with Locally Linear Embedding 119
4.1.1.3 Evaluation Criterion for Parameter Selection 120
4.1.1.4 Results and Analysis 122
4.1.2 Statistical Analysis of Tongue Color Distribution 124
4.1.2.1 Tongue Color Gamut: Generation and Modeling 125
4.1.2.2 Tongue Color Centers 134
4.1.2.3 Distribution of Typical Image Features 137
4.1.2.4 Separation of Tongue Coating and Substance 141
4.2 Tongue Texture Analysis 143
4.3 Tongue Shape Analysis 144
4.3.1 Shape Correction 144
4.3.1.1 Automatic Contour Extraction 145
4.3.1.2 The Length Criterion 145
4.3.1.3 The Area Criterion 146
4.3.1.4 The Angle Criterion 147
4.3.1.5 Correction by Combination 148
4.3.2 Extraction of Shape Features 149
4.3.2.1 The Length-based Feature 149
4.3.2.2 The Area-based Feature 151
4.3.2.3 The Angle-based Feature 153
4.3.3 Tongue Shape Classification 153
4.3.3.1 Modeling The Classification as A Hierarchy 155
4.3.3.2 Calculating Relative Weights 156
4.3.3.3 Calculating The Global Weights 157
4.3.3.4 Fuzzy Shape Classification 157
4.4 Extraction of Other Local Pathological Features 158
4.4.1 Petechia 158
4.4.2 Tongue Crack 160
4.4.3 Tongueprint 160
4.4.4 Sublingual Veins 161
4.5 Summary 162
References\r 163
Chapter 5 Computerized Tongue Diagnosis 167
5.1 Bayesian Network for Computerized Tongue Diagnosis 167
5.1.1 Quantitative Pathological Features Extraction 167
5.1.2 Bayesian Networks 169
5.1.3 Experimental Results 171
5.1.3.1 Bayesian Network Classifier Based on Textural Features 174
5.1.3.2 Bayesian Network Classifier Based on Chromatic Features 174
5.1.3.3 Bayesian Network Classifier Based on Combined Features 176
5.2 Diagnosis Based on Hyperspectral Tongue Images 178
5.2.1 Hyperspectral Tongue Images 179
5.2.2 The SVM Classifier Applied to Hyperspectral Tongue Images 180
5.2.2.1 Linear SVM: Linearly Separable 180
5.2.2.2 Linear SVM: Linearly Non-Separable 181
5.2.2.3 Non-linear SVM 182
5.2.3 Experimental Results 183
5.2.3.1 Experiment 1: Comparing Linear and Non-Linear SVM, RBFNN and K-NN Classifiers 183
5.2.3.2 Experiment 2: Evaluating the Diagnostic Performance of SVM 184
5.3 Summary 186
References\r 187
PART III: COMPUTERIZED PULSE SIGNAL ANALYSIS 189
Chapter 6 Pulse Signal Acquisition and Preprocessing 191
6.1 Pressure Pulse Signal Acquisition 191
6.1.1 Application Scenario and Requirement Analysis 192
6.1.2 System Architecture 193
6.1.2.1 Mechanical Structure 194
6.1.2.2 Sensor 196
6.1.2.3 Circuit 199
6.1.2.4 Software Architecture 201
6.1.3 Multi-Channel Pulse Signals 201
6.2 Baseline Wander Correction of Pulse Signals 206
6.2.1 Detecting the Onsets of Pulse Wave 207
6.2.2 Wavelet Based Cascaded Adaptive Filter 209
6.2.2.1 The Design of CAF Filter 209
6.2.2.2 Detection the Level of Baseline Wander Using ER 212
6.2.2.3 The Discrete Meyer Wavelet Filter 216
6.2.2.4 Cubic Spline Estimation Filter 219
6.2.3 Results on Actual Pulse Signals 221
6.3 Summary 223
References 224
Chapter 7 Feature Extraction of Pulse Signals 227
7.1 Spatial Feature Extraction 227
7.1.1 Fiducial Point-based Methods 227
7.1.2 Approximate Entropy 229
7.2 Frequency Feature Extraction 230
7.2.1 Hilbert-Huang Transform 230
7.2.2 Wavelet and Wavelet Packet Transform 232
7.2.2.1 Wavelet Transform 233
7.2.2.2 Wavelet Packet Transform 233
7.3 AR Model 234
7.4 Gaussian Mixture Model 236
7.4.1 Two-term Gaussian Model 236
7.4.2 Feature Selection 240
7.4.3 FCM Clustering 242
7.5 Summary 242
References 243
Chapter 8 Classification of Pulse Signals 245
8.1 Pulse Waveform Classification 245
8.1.1 Modules of Pulse Waveform Classification 246
8.1.1.1 Digital Pulse Waveform Acquisition 246
8.1.1.2 Pulse Waveform Preprocessing 248
8.1.1.3 Pulse Waveform Feature Extraction and Classification 249
8.1.2 The EDFC and GEKC Classifiers 251
8.1.2.1 ERP, DFWKNN, and KDFWKNN 251
8.1.2.2 EDFC and GEKC 252
8.1.3 Experimental Results 255
8.2 Arrhythmic Pulses Detection 257
8.2.1 Clinical Value of Pulse Rhythm Analysis 257
8.2.2 Automatic Recognition of Pulse Rhythms 259
8.2.2.1 Lempel-Ziv Complexity Analysis 259
8.2.2.2 Definitions and Basic Facts 260
8.2.2.3 Automatic Recognition of Pulse Patterns Distinctive in Rhythm 263
8.2.3 Experimental Results 272
8.3 Combination of Heterogeneous Features for Pulse Diagnosis 274
8.3.1 Multiple Kernel Learning 275
8.3.1.1 Kernel Functions 276
8.3.1.2 SimpleMKL 277
8.3.2 Experimental Results and Discussion 279
8.3.2.1 Classification Experimental of Wrist Blood Flow Signal 279
8.3.2.2 Pulse Waveform Classification 282
8.4 Summary 282
References 283
PART IV: COMPUTERIZED ODOR SIGNAL ANALYSIS 287
Chapter 9 Breath Analysis System: Design and Optimization 289
9.1 Breath Analysis 289
9.2 Design of Breath Analysis System 291
9.2.1 Description of the System 291
9.2.1.1 Breath Gas Collecting 292
9.2.1.2 Signal Sampling 293
9.2.2 Signal Sampling and Preprocessing 296
9.2.2.1 Sampling Procedure 297
9.2.2.2 Preprocessing 298
9.3 Sensor Selection 299
9.3.1 Linear Discriminant Analysis 299
9.3.1.1 Data Expression 300
9.3.1.2 The Optimum Direction of LDA 301
9.3.1.3 Difference Between Two Classes as the Linear Combination of Sensors 302
9.3.1.4 Weight of Sensor 304
9.3.1.5 Algorithm 304
9.3.2 Sensor Selection in Breath Analysis System 304
9.3.2.1 Sensor Selection for Disease Diagnosis 305
9.3.2.2 Sensor Selection for Evaluating the Medical Treatment 311
9.3.3 Comparison Experiment and Performance Analysis 314
9.3.3.1 Sensor Selection for Disease Diagnosis 314
9.3.3.2 Sensor Selection for Evaluating the Medical Treatment 316
9.4 Summary 317
References 317
Chapter 10 Feature Extraction and Classification of Breath Odor Signals 321
10.1 Feature Extraction of Odor Signals 321
10.1.1 Geometry Features 322
10.1.2 Principal Component Analysis 324
10.1.3 Wavelet Packet Decomposition 324
10.1.4 Gaussian Function Representation 325
10.1.5 Gaussian Basis Representation 331
10.1.6 Experimental Results 334
10.1.6.1 Experiment I: Decide Which Method is the Best 334
10.1.6.2 Experiment II: Test for GBR 336
10.2 Common Classifiers for Odor Signal Classification 336
10.2.1 K Nearest Neighbor 337
10.2.2 Artificial Neural Network 337
10.2.3 Support Vector Machine 337
10.3 Sparse Representation Classification 338
10.3.1 Data Expression 338
10.3.2 Test Sample Representation by Training Samples 339
10.3.3 Samples Sampling Errors 340
10.3.4 Voting Rules 341
10.3.4.1 Coefficients Based Voting Rule 341
10.3.4.2 Residue Based Voting Rule 341
10.3.5 Identification Steps 342
10.4 Support Vector Ordinal Regression 342
10.4.1 Problem Analysis 342
10.4.2 Basic Idea of Support Vector Regression 343
10.4.3 Support Vector Ordinal Regression 344
10.4.4 The Dual Problem 345
10.4.5 Identification Steps 346
10.5 Evaluation on Classification methods 347
10.5.1 Evaluation on SRC 347
10.5.2 Evaluation on SRC 351
10.6 Summary 355
References 355
Index 359