Advanced Machine Vision Paradigms for Medical Image Analysis

Description

Computer vision and machine intelligence paradigms are prominent in the domain of medical image applications, including computer assisted diagnosis, image guided radiation therapy, landmark detection, imaging genomics, and brain connectomics. Medical image analysis and understanding are daunting tasks owing to the massive influx of multi-modal medical image data generated during routine clinal practice. Advanced computer vision and machine intelligence approaches have been employed in recent years in the field of image processing and computer vision. However, due to the unstructured nature of medical imaging data and the volume of data produced during routine clinical processes, the applicability of these meta-heuristic algorithms remains to be investigated.

Advanced Machine Vision Paradigms for Medical Image Analysis presents an overview of how medical imaging data can be analyzed to provide better diagnosis and treatment of disease. Computer vision techniques can explore texture, shape, contour and prior knowledge along with contextual information, from image sequence and 3D/4D information which helps with better human understanding. Many powerful tools have been developed through image segmentation, machine learning, pattern classification, tracking, and reconstruction to surface much needed quantitative information not easily available through the analysis of trained human specialists. The aim of the book is for medical imaging professionals to acquire and interpret the data, and for computer vision professionals to learn how to provide enhanced medical information by using computer vision techniques. The ultimate objective is to benefit patients without adding to already high healthcare costs.

Explores major emerging trends in technology which are supporting the current advancement of medical image analysis with the help of computational intelligence
Highlights the advancement of conventional approaches in the field of medical image processing
Investigates novel techniques and reviews the state-of-the-art in the areas of machine learning, computer vision, soft computing techniques, as well as their applications in medical image analysis

Front Cover 2
ADVANCED MACHINE VISION PARADIGMS FOR MEDICAL IMAGE ANALYSIS 3
ADVANCED MACHINE VISION PARADIGMS FOR MEDICAL IMAGE ANALYSIS 5
Copyright 6
Dedication 7
Contents 9
Contributors 13
Preface 17
1 - An introductory illustration of medical image analysis 21
- 1. Introduction 21
- 2. Medical image analysis: issues and challenges 22
- 3. Summary of all contributory chapters (challenges and findings) 25
- 4. Discussion and conclusion with future trends 28
- References 28
2 - Computer-aided decision support system for symmetry-based prenatal congenital heart defects 31

Front Cover 2
ADVANCED MACHINE VISION PARADIGMS FOR MEDICAL IMAGE ANALYSIS 3
ADVANCED MACHINE VISION PARADIGMS FOR MEDICAL IMAGE ANALYSIS 5
Copyright 6
Dedication 7
Contents 9
Contributors 13
Preface 17
1 - An introductory illustration of medical image analysis 21
- 1. Introduction 21
- 2. Medical image analysis: issues and challenges 22
- 3. Summary of all contributory chapters (challenges and findings) 25
- 4. Discussion and conclusion with future trends 28
- References 28
2 - Computer-aided decision support system for symmetry-based prenatal congenital heart defects 31
- 1. Introduction 31
- 2. Clinical dataset, clinical perspectives of CHD screening and methodology 34
  - 2.1 Dataset collection and database development 34
  - 2.2 Clinical background of ultrasonography in asymmetric heart-oriented CHD screening 36
  - 2.3 Methodology 38
    - 2.3.1 Fuzzy inference rule-based adaptive maximum likelihood estimation-based preprocessing 39
    - 2.3.2 Morphological operations 41
      - 2.3.2.1 Entropy thresholding 41
      - 2.3.2.2 Morphological processing 45
      - 2.3.2.3 Feature extraction 47
    - 2.3.3 ANFIS classifier 50
  - 3. Results and discussion 53
  - 4. Conclusion 70
  - References 70
- 3 - Morphological extreme learning machines applied to the detection and classification of mammary lesions 75
  - 1. Introduction 76
    - 1.1 Main goals 77
    - 1.2 Chapter structure 78
  - 2. Breast cancer 78
    - 2.1 Breasts and cancer classification in mammography 79
  - 3. Extreme learning machines 81
    - 3.1 Morphological operators 83
    - 3.2 Image descriptors 86
      - 3.2.1 Haralick features 87
      - 3.2.2 Wavelet series decomposition 87
    - 3.3 Metrics 89
  - 4. Proposed methodology 89
    - 4.1 Morphological approximation in binary images 89
    - 4.2 Proposal: morphological neurons 92
    - 4.3 Detection and classification of lesion in mammograms 94
    - 4.5 Classification using the proposed method (morphological extreme learning machine) 95
    - 4.6 Image Retrieval in Medical Applications database 95
    - 4.4 Image feature extraction 95
  - 5. Results 98
    - 5.1 Results and experimental analysis 101
  - 6. Conclusion 111
    - 6.1 Contributions 112
    - 6.2 Future study 112
  - References 113
- 4 - 4D medical image analysis: a systematic study on applications, challenges, and future research directions 117
  - 1. Introduction 118
  - 2. Different 4D medical imaging modalities 119
    - 2.1 4D computer tomography 119
    - 2.2 4D positron emission tomography scan 119
    - 2.3 4D Ultrasound 120
    - 2.4 4D MRI 121
  - 3. Exploration of 4D CT 122
    - 3.1 Application domains of 4D CT medical imaging 122
      - 3.1.2 Radiation oncology 123
      - 3.1.1 Parathyroid 123
      - 3.1.5 Cardiac 125
      - 3.1.3 Wrist 125
      - 3.1.4 Hip 125
    - 3.2 Challenges in 4D CT 126
  - 4. Exploration of 4D ultrasound imaging 126
    - 4.1 4D ultrasound 126
      - 4.1.1 Evolution from 2D to 4D 127
      - 4.1.2 Steps in ultrasound image capturing 128
    - 4.2 4D ultrasound imaging in the assessment of fetal behavior 130
    - 4.3 4D ultrasound imaging in neuroophthalmology 132
    - 4.4 4D ultrasound imaging in neurology 133
    - 4.5 Role of ultrasound in medical field 134
    - 4.6 Challenges in ultrasound imaging 135
  - 5. Exploration of 4D flow magnetic resonance imaging 135
    - 5.1 4D flow MRI 135
    - 5.2 4D flow MRI Processing and Analysis 136
      - 5.2.1 Data acquisition using flow-sensitive MRI 137
      - 5.2.2 Data Preprocessing for Error Corrections 139
    - 5.3 4D blood flow visualization and Quantification 140
    - 5.4 Applications of 4D flow MRI 141
      - 5.4.2 Vessel segmentation in pancreas 141
      - 5.4.1 Evaluation of left ventricular diastolic function 141
      - 5.4.4 Assessment of uterine artery geometry and hemodynamics in human pregnancy 142
      - 5.4.3 Longitudinal monitoring of hepatic blood flow before and after TIPS 142
      - 5.4.5 Future of 4D flow MRI 143
    - 6. Machine learning techniques in 4D medical imaging 143
      - 6.1 Machine learning approaches in 4D CT 143
      - 6.2 Machine learning approaches in 4D ultrasound 144
      - 6.3 Machine learning approaches in 4D flow MRI 144
    - 7. Conclusion and future trends 146
    - References 146
  - 5 - Comparative analysis of hybrid fusion algorithms using neurocysticercosis, neoplastic, Alzheimer's, and as ... 151
    - 1. Introduction 152
      - 1.1 Levels of medical image fusion 153
        
        1.1.1 Pixel-level fusion 153
        
        1.1.2 Feature-level fusion 153
        
        1.1.3 Decision-level fusion 154
      - 1.2 Multimodal medical images 154
        
        1.2.2 Magnetic resonance imaging 154
        
        1.2.3 Positron emission tomography 154
        
        1.2.1 Computed tomography 154
        
        1.2.4 Single photon emission computed tomography 155
      - 1.3 Medical image fusion techniques 155
        
        1.3.3 MRI-CT fusion 155
        
        1.3.1 MRI-SPECT fusion 155
        
        1.3.2 MRI-PET fusion 155
      - 1.4 Dataset descriptions 155
        
        1.4.1 Neurocysticercosis 156
        
        1.4.2 Neoplastic disease 156
        
        1.4.4 Alzheimer's disease 157
        
        1.4.3 Astrocytoma 157
      - 1.5 Categories of multimodality images 158
    - 2. Literature survey 160
    - 3. Proposed work 161
      - 3.1 Generalized MIF scheme 161
        
        3.1.1 Input image modalities 162
        
        3.1.2 Medical image registration 162
        
        3.1.3 Fusion process 162
        
        3.1.4 Evaluation [17] 163
        
        3.1.4.1 Performance evaluation metrics 163
        
        3.2 Proposed hybrid fusion algorithms 166
        
        3.2.3 Hybrid Fusion algorithm Based on dfDFRWT-DTCWT 166
        
        3.2.1 Hybrid Fusion Algorithm Based on Dual Tree Complex Wavelet Transform (DTCWT)-sNSST 166
        
        3.2.2 Hybrid Fusion algorithm Based on Nonssubsampled Contourlet Transform (NSCT)-sNSST 166
        
        3.2.4 Hybrid Fusion algorithm Based on sNSCT-dfDFRWT 169
        
        3.2.5 Hybrid Fusion algorithm Based on sNSCT-DTCWT 169
        
        4. Implementations and discussions 172
        
        5. Summary 186
        
        References 186
      - 6 - Binary descriptor design for the automatic detection of coronary arteries using metaheuristics 189
        
        1. Introduction 189
        
        2. Background 192
        
        2.1 Morphological top-hat transform 192
        
        2.2 Iterated local search 192
        
        2.3 Univariate marginal distribution algorithm 195
        
        2.4 Genetic algorithms 197
        
        3. Proposed method 198
        
        3.1 Design of binary descriptors using Metaheuristics 198
        
        3.2 Evaluation measures 200
        
        4. Computational experiments 200
        
        4.1 Results of vessel enhancement 201
        
        4.2 Results of vessel segmentation 204
        
        5. Concluding remarks 204
        
        Appendix A 205
        
        References 207
        
        ACKNOWLEDGMENTS 207
      - 7 - A cognitive perception on content-based image retrieval using an advanced soft computing paradigm 209
        
        1. Introduction 210
        
        2. Related works 211
        
        3. Methodology 213
        
        3.1 MRI brain image database 214
        
        3.2 Pre-processing 214
        
        3.2.1 Median filtering 214
        
        3.3 Feature extraction 215
        
        3.3.1 Gray-level cooccurrence matrix 215
        
        3.3.2 Local binary patterns 217
        
        3.3.2.1 Calculation of local binary pattern 217
        
        3.3.3 Moment invariant feature 218
        
        3.4 Feature selection 219
        
        3.4.1 Differential evolution 219
        
        3.5 Feature vector 219
        
        3.6 Feature subset matrix 220
        
        3.7 SVM classifier 220
        
        3.7.1 Algorithm 221
        
        3.7.1.1 Separable case 221
        
        3.7.1.2 Nonseparable case 222
        
        3.8 Feature space and construction 222
        
        3.9 Query feature extraction 223
        
        3.11 Retrieval 224
        
        3.10 Euclidean search 224
        
        4. Experimental results 224
        
        5. Conclusion 230
        
        References 230
        
        8 - Early detection of Parkinson's disease using data mining techniques from multimodal clinical data 233
        
        1. Introduction 233
        
        2. Review of literature 234
        
        3. Materials and methods 235
        
        4. Voice dataset processing 236
        
        4.1 Clustering voice dataset using K-means 237
        
        4.2 Voice dataset classification using decision tree 238
        
        4.3 Voice dataset classification using SVM 238
        
        4.5 Voice dataset classification using the SVM and KNN 240
        
        4.4 Voice dataset classification using random forest 240
        
        5. Spiral dataset preprocessing 240
        
        6. Ensemble-based prediction from voice and spiral dataset 242
        
        7. Evaluation metrics 243
        
        8. Experimental results 243
        
        9. Conclusion 244
        
        References 247
        
        9 - Contrast improvement of medical images using advanced fuzzy logic-based technique 249
        
        1. Introduction 249
        
        2. Review of the recent works and motivation toward the proposed work 250
        
        3. Steps involved in the proposed approach 255
        
        4. Methods and materials 256
        
        4.1 Medical imaging 256
        
        4.2 Noise and its types 257
        
        4.3 Improved mean value-based median filter, Improved_Median() 258
        
        4.5 Type-2 fuzzy set 260
        
        4.4 HSV color space 260
        
        4.6 Type-2 (V-channel) 262
        
        4.7 Adaptive ameliorated CLAHE, AA_CLAHE() 263
        
        5. Experiment and result discussion 265
        
        7. Objective evaluation 270
        
        6. Subjective evaluations 270
        
        8. Time comparison 274
        
        9. Conclusion 274
        
        References 275
        
        10 - Bone age assessment using metric learning on small dataset of hand radiographs 279
        
        1. Introduction 279
        
        2. Motivation 280
        
        3. Related work 281
        
        4. Methodology 282
        
        4.1 Data 282
        
        4.2 Architecture 283
        
        4.3 Loss function 285
        
        4.4 Embedding function 286
        
        5. Results 287
        
        5.1 Training details and hyperparameters 287
        
        5.2 Visualizing embeddings 287
        
        6. Discussion 288
        
        7. Conclusion 290
        
        References 290
        
        11 - Conclusion and future research directions 293
        
        1. Concluding remarks 293
        
        References 296
        
        2. Future avenues 296
        
        Index 299
        
        A 299
        
        B 299
        
        C 300
        
        D 301
        
        E 302
        
        F 302
        
        G 303
        
        I 304
        
        H 304
        
        L 305
        
        J 305
        
        M 305
        
        K 305
        
        N 306
        
        Q 307
        
        O 307
        
        R 307
        
        P 307
        
        T 308
        
        S 308
        
        W 309
        
        U 309
        
        X 309
        
        V 309
        
        Back Cover 311

Author's affiliation

Siddhartha Bhattacharyya: RCC Institute of Information Technology, Kolk

Description

TOC

Reviews (0)

Author's affiliation