Medical Imaging and Computer-Aided Diagnosis: Proceedings of 2022 International Conference on Medical Imaging and Computer-Aided Diagnosis (MICAD 2022) (Lecture Notes in Electrical Engineering, 810)

This book covers virtually all aspects of image formation in medical imaging, including systems based on ionizing radiation (x-rays, gamma rays) and non-ionizing techniques (ultrasound, optical, thermal, magnetic resonance, and magnetic particle imaging) alike. In addition, it discusses the development and application of computer-aided detection and diagnosis (CAD) systems in medical imaging. Given its coverage, the book provides both a forum and valuable resource for researchers involved in image formation, experimental methods, image performance, segmentation, pattern recognition, feature extraction, classifier design, machine learning / deep learning, radiomics, CAD workstation design, human–computer interaction, databases, and performance evaluation. Organization Preface Contents Medical Imaging Optimizing the Non-local Means Filtering of CT Images 1 Introduction 2 Algorithm Description 2.1 The Non-local Means Filter 2.2 Proposed Optimization Procedure 3 Experimental Results 4 Discussion 5 Conclusions References Exploring Structure-Wise Uncertainty for 3D Medical Image Segmentation 1 Introduction 2 Related Work 3 Methods 3.1 Structure Definition 3.2 Uncertainty Estimation Methods 3.3 Uncertainty Aggregation Techniques 4 Experiments 4.1 Data 4.2 Experimental Setup 4.3 Results 4.4 Discussion 5 Conclusion References Towards Developing a Lightweight Neural Network for Liver CT Segmentation 1 Introduction 2 Proposed Methodology 2.1 Network Architecture 3 Setup for the Experiment 3.1 Data 3.2 Implementation Details 4 Results 5 Conclusion References NuRISC: Nuclei Radial Instance Segmentation and Classification 1 Introduction 2 Related Work 3 Methodology 3.1 Deep Regression Network 3.2 Instance Representation 3.3 Radial Distance Maps 3.4 Pre-processing 3.5 Model Architecture 3.6 Loss Functions 3.7 Post-processing 4 Experiments and Results 4.1 Implementation Details 4.2 Evaluation Metrics 4.3 Datasets 4.4 Baseline Methods 4.5 Experimental Results 5 Conclusion References A Semi-supervised Framework for Automatic Pixel-Wise Breast Cancer Grading of Histological Images 1 Introduction 2 Related Work 3 Method 3.1 Semi-supervised Learning Framework Based on EM Model 3.2 Patch Selection 4 Experimental Results 4.1 Dataset for Training and Validation 4.2 Data Preprocess 4.3 Patch Classification 4.4 Pixel-Wise Classification on WSI 4.5 FROC Acceptance 5 Conclusion References Lunatum Prosthetic Replacement: Modeling Based on Volume Rendering of CT Scan Images 1 Introduction 2 Lunatum Anatomy and Associated Problems 3 Biomaterial Selection 4 Material and Methods 4.1 Data Acquisition 4.2 Segmentation 4.3 3D Model Creation Based on Volume Rendering 4.4 3D Printing (Fused Disposition Modeling FDM) 5 Results 6 Conclusion References Augmented Reality Applications for Image-Guided Robotic Interventions Using Deep Learning Algorithms 1 Introduction 2 Related Work 2.1 Medical Image Registration Based on Deep Learning 2.2 Augmented Reality Based on DL Image Registration 3 Methodology 3.1 Dataset and Implementation Details 3.2 Evaluation Metrics 3.3 Experiment Results 4 Discussion and Conclusions References Transfer Learning Based Classification of Diabetic Retinopathy on the Kaggle EyePACS Dataset 1 Introduction 2 Related Work 3 Pre-trained Models 4 Dataset 5 Methodology 5.1 Transfer Learning Details 5.2 Training Using Pre-trained Models 6 Discussion 7 Conclusion and Future Work References Ex-vivo Evaluation of Newly Formed Bone After Lumbar Interbody Fusion Surgery Using X-ray Micro Computed Tomography 1 Introduction 2 MicroCT Bone Tissue Evaluation 3 Materials and Methods 3.1 Datasets Used 3.2 Determination of Volume of Interest 3.3 Image Analysis 4 Results 5 Newly Formed Bone Evaluation 6 Conclusion and Future Work References Community Detection in Medical Image Datasets: Using Wavelets and Spectral Methods 1 Introduction 2 Our Method 3 Results 4 Inferring the Disease Spectrum 5 Conclusions References Non-pooling Network for Medical Image Segmentation 1 Introduction 2 Methodology 2.1 Basic Block 2.2 Attention Enhancement Module 2.3 Feature Enhancement Module 3 Experiments and Results 3.1 Datasets 3.2 Experimental Settings 3.3 Experimental Results 4 Conclusion References Lung CT Analysis Using 3D Disparity-Regularised Block Matching for Stereotactic Ablative Body Radiotherapy 1 Introduction 2 Methodology 2.1 Materials and Data Preprocessing 2.2 Lung CT Analysis Using Extended DBLM 2.3 Parameters of Extended DBLM and Its Performance Evaluation 3 Results and Discussion 4 Conclusion References Identification of Melanoma Diseases from Multispectral Dermatological Images Using a Novel BSS Approach 1 Introduction 2 Proposed Method 3 Results 4 Conclusion References 2.5D Lightweight Network Integrating Multi-scale Semantic Features for Liver Tumor Segmentation 1 Introduction 2 Related Work 2.1 Inception Architecture 2.2 Residual Network 2.3 2.5D Network Architecture 3 Method 3.1 MAA_Net 3.2 Encoder-Decoder Structure 3.3 Dual-Feature Fusion Module 3.4 Loss Function 4 Experiments 4.1 Datasets 4.2 Setting Details 4.3 Results and Analysis 5 Conclusion References Registration of Medical Image Sequences Using Auto-differentiation 1 Introduction 2 Methods 2.1 Image Sequence Registration 2.2 Experimental Data 2.3 Implementation Details 3 Results and Discussion 4 Conclusions References Small Animal Imaging: Iterative Algorithms Combined with Regularization Schemes, an Application to a Dual-Head Small Animal PET 1 Introduction 2 Theory 3 Materials and Methods 4 Results and Discussion 5 Conclusions References Early Detection of Parkinson's Disease Dementia Using Dual-Sided Multi-scale Convolutional Neural Networks (DSMS-CNN) 1 Introduction 2 Proposed Method 2.1 Region of Interest 2.2 Multi-scale Convolutional Neural Network 2.3 Proposed Dual-Sided Architecture 3 Experiments and Analysis 3.1 Dataset 3.2 Preprocessing 3.3 Parameter Settings 3.4 Ablation Study 3.5 Comparison with State-of-the-Art 4 Conclusions References A Change Detection with Machine Learning Approach for Medical Image Analysis 1 Introduction 1.1 Basics of Change Detection 1.2 Deterministic Methods 1.3 Probabilistic Methods 2 Change Detection Applied to Biomedical Imaging 3 Related Work 4 Case Study 4.1 Numerical Algorithms 4.2 Spatial Algorithms 4.3 Connectivity Indices 4.4 Feature Selection 4.5 Factorial Analysis (PCA) 4.6 Results of the Proposed Methodology 5 Conclusion References U-Net##: A Powerful Novel Architecture for Medical Image Segmentation 1 Introduction 2 The Proposed Architecture 2.1 Parallel Neural Networks 2.2 Auxiliary Convolutional Blocks 2.3 Deep Supervision 3 Experiments and Results 3.1 Dataset and Pre-processing 3.2 Implementation Details 3.3 Results 4 Conclusions References Computer-Aided Detection/Diagnosis Optimising Chest X-Rays for Image Analysis by Identifying and Removing Confounding Factors 1 Introduction 2 Materials and Methods 3 Experimental Analysis and Results 4 Discussion and Conclusion References 3D-3D Rigid Registration: A Comparative Analysis Study on Femoral Bone Scans 1 Introduction 2 Registration Algorithms 3 Material and Methods 3.1 Material and Preprocessing 3.2 Simulation Workflow 3.3 Preprocessing 4 Experiments and Results 4.1 Parametrization of the Registration Algorithms 4.2 Registration Time 4.3 Convergence 4.4 Registration Accuracy 5 Conclusion and Future Work References Fully Automatic Axial Vertebral Rotation Measurement of Children with Scoliosis Using Convolutional Neural Networks 1 Introduction 2 Methodology 2.1 Proposed Method 2.2 Validation 3 Results and Discussion 3.1 Spinal Feature Segmentation 3.2 AVR Measurement 4 Conclusion References Diagnostic Accuracy and Reliability of Deep Learning-Based Human Papillomavirus Status Prediction in Oropharyngeal Cancer 1 Introduction 2 Methods 2.1 Study Cohort 2.2 CT Pre-processing 2.3 Deep Neural Network: Architecture and Training 2.4 Reliability Assessment 3 Results 3.1 Study Cohort 3.2 Diagnostic Accuracy and Reliability Assessment 4 Discussion 5 Conclusion References Optimizing the Illumination of a Surgical Site in New Autonomous Module-based Surgical Lighting Systems 1 Introduction 2 Related Work 3 Implementation 3.1 Occluder Types 3.2 Representation of the Surgical Site 3.3 Optimization Pipeline 3.4 Light Intensity Optimization 4 Results 4.1 Methods 4.2 Shadow Reduction 4.3 Temporal Brightness Distribution 5 Conclusions and Future Works References An Eye-Tracking Based Machine Learning Model Towards the Prediction of Visual Expertise for Electrocardiogram Interpretation 1 Introduction 1.1 Background 1.2 Related Works 2 The Dataset 2.1 Overview 2.2 Data Collection Method 2.3 Ethics 3 Data Processing 4 Results 5 Conclusion References Synthetic Data as a Tool to Combat Racial Bias in Medical AI: Utilizing Generative Models for Optimizing Early Detection of Melanoma in Fitzpatrick Skin Types IV–VI 1 Introduction 2 Background 2.1 Medical Examination 2.2 CAD Examination 3 Related Works 3.1 Generative Adversarial Networks 3.2 Zero-Shot Text-to-Image Generative Models 4 Experiments and Results 4.1 Dermatological Perspective 4.2 CAD Perspective 5 Conclusions References BD-Transformer: A Transformer-Based Approach for Bipolar Disorder Classification Using Audio 1 Introduction 2 Related Work 3 Proposed Framework 3.1 Preprocessing 3.2 Feature Extraction 4 Experiments and Results 4.1 Dataset 4.2 Implementation Details 4.3 Performance of the Proposed Approach for Bipolar Disorder Diagnosis 4.4 Comparison with State of the Art Methods 5 Conclusion and Future Work References Establishment and Analysis of a Combined Diagnostic Model of Acute Myocardial Infarction Based on Random Forests and Artificial Neural Networks 1 Introduction 2 Materials and Methods 2.1 Data Collection 2.2 Differential Expression and Enrichment Analysis 2.3 Random Forest Screening for the Important Genes 2.4 PPI Network Analysis 2.5 Neural Network for Building the Disease Classification Model 2.6 Additional Data Verification 3 Results 3.1 Differential Expression Analysis 3.2 GO/KEGG Enrichment Analysis 3.3 Random Forest Screening for DEGs 3.4 Protein–Protein Interaction (PPI) Network Analysis 3.5 Construction of the ANN Model 3.6 Evaluation of the AUC 4 Discussion References Striped-Cross Attention Network with Implicit Semantic Knowledge for Antibody Structure Prediction 1 Introduction 2 Methodology 2.1 The Diagram of Striped-Cross Attention 2.2 Classifier Network Structure Based on Striped-Cross Attention 2.3 Framework Architecture for Antibody Structure Prediction 3 Experiment 3.1 Dataset 3.2 Evaluation of Inter-Residue Distances and Orientations Prediction 3.3 Evaluation of the Antibody Structure Prediction 4 Conclusion References A Mobile Monitoring Application for Post-traumatic Stress Disorder 1 Introduction 2 Related Work 2.1 Mobile and Web Applications for Health Monitoring and Surveillance 2.2 Mental Disorders and Mental Health Telemedicine 2.3 Mobile Application for Post-traumatic Stress Disorder Diagnosis and Follow-Up 3 The Mobile Application 3.1 Video and Data Acquisition 3.2 Virtual Interviewer 3.3 PCL-5 Questionnaire 3.4 Patient History Recording 4 Application Development 4.1 General Architecture 4.2 Technological Choice 5 Application Evaluation and Dataset Collection 6 Conclusions and Future Work References COVID-19 Diagnosis and Classification from CXR Images Using Vision Transformer 1 Introduction 2 Related Work 3 Proposed Model 4 Data 4.1 Dataset Details 4.2 Data Preprocessing 5 Experiments and Analysis 5.1 Implementation Details 5.2 Evaluation Method 5.3 Results and Discussion 6 Conclusion References Improved Techniques for the Conditional Generative Augmentation of Clinical Audio Data 1 Introduction 2 Materials and Methods 2.1 Data Set, Preprocessing, and Benchmark Augmentations 2.2 Proposed Data Augmentation Method 2.3 Classifier for Evaluation 3 Results 4 Discussion 5 Conclusion References Learning from Failure: A Methodology for the Retrieve Stage of a Cardiovascular Case-Based Reasoning System 1 Introduction 2 Used Strategies to Retrieve Cases 3 Other Measures Beyond Similarity 3.1 Similarity 3.2 Index Variation 3.3 Success Ratio 4 Evaluating Case Retrieval Strategies 5 Conclusions and Future Directions References Machine Learning and Deep Learning Forming of Validation Dataset for Deep Learning Based Model of Medical Image Grouping 1 Introduction 2 Challenges in Selection of the Valuable Source of Medical Images 3 Forming of Distinctive Groups, Constrained Combinations 4 Process of Grouping and Algorithms 5 Results 6 Conclusion References Deep Learning Based Radiomics to Predict Treatment Response Using Multi-datasets 1 Introduction 2 Entropy 2.1 Havrda-Charvat Cross-Entropy 3 Neural Network Architecture for Relapse Prediction 4 Experimentations 4.1 Datasets 4.2 Evaluation Method 4.3 Results 5 Discussion 6 Conclusion References Convolutional Neural Network Classification of Liver Fibrosis Stages Using Ultrasonic Images Colorized by Features of Echo-Envelope Statistics 1 Introduction 2 Method 2.1 Dataset 2.2 Formation and Selection of Input Images 2.3 Modulation and Colorization of Input Images 2.4 Learning and Validation of Networks 3 Result and Discussion 4 Conclusion References FedRNN: Federated Learning with RNN-Based Aggregation on Pancreas Segmentation 1 Introduction 2 Materials and Methods 2.1 Dataset 2.2 Federated Learning Framework 2.3 Model Aggregation 2.4 Experimental Setup 3 Results 4 Discussion 5 Conclusion References UNet-2022: Exploring Dynamics in Non-isomorphic Architecture 1 Introduction 2 Related Work 3 Methodology 3.1 Parallel Non-isomorphic Block 3.2 Convolution/De-convolution Stem 4 Experiments 4.1 Dataset 4.2 Implementation Details 4.3 Comparisons on Abdominal Multi-organ Segmentation 4.4 Comparisons on Automated Cardiac Diagnosis 4.5 Comparisons on Neural Structures Segmentation 4.6 Comparisons on Skin Lesion Segmentation 4.7 Ablation Studies of Modules and Strategies 5 Conclusion References Hybrid-Fusion Transformer for Multisequence MRI 1 Introduction 2 Method 2.1 Hybrid Fusion from CNN Encoders 2.2 CNN Decoder and Loss Function 3 Experiments 3.1 Datasets 3.2 Quantitative Results 3.3 Qualitative Results 4 Discussion 5 Conclusion References STResNet: Covid-19 Detection by ResNet Transfer Learning and Stochastic Pooling 1 Introduction 2 Dataset 3 Methodology 3.1 ResNet-50 3.2 Stochastic Pooling 3.3 Support Vector Machine 4 Experiment Results and Discussion 4.1 Experiment Results of STResNet 4.2 Data Augmentation Results 4.3 Stochastic Pooling Against Max Pooling and Average Pooling 4.4 Comparison to State-to-the-Art Approaches 5 Conclusion References Convolutional Neural Networks for Newborn Pain Assessment Using Face Images: A Quantitative and Qualitative Comparison 1 Introduction 2 Materials and Methods 2.1 Face Images Datasets 2.2 Classification Models 2.3 Training/Test Protocol 2.4 Explainable Artificial Intelligence 3 Results and Discussion 3.1 Quantitative Results 3.2 Qualitative Results 3.3 Discussion 4 Conclusion References Machine Learning for the Evaluation and Detection of Key Markers in Dilated Cardiomyopathy 1 Introduction 2 Methods and Materials 2.1 Data Acquisition and Download 2.2 Data Processing and Genetic Screening 2.3 Enrichment Analysis: GO, KEGG, DO and GSEA 2.4 Selecting and Identifying Gene Predictor Models for Premature Diagnosis 2.5 Immune Cell Infiltration Analysis 3 Results 3.1 Screening for DCM-Associated Differential Genes 3.2 Co-expressed Genes GO, DO and KEGG Signalling Path Enrichment Analysis 3.3 Functional Clustering of DCM-GSEA Analysis 3.4 Screening and Identification of Gene Prediction Models for Early Diagnosis 3.5 The Immune Checkpoint Related Genes Analysis 4 Discussion 5 Conclusion References Others Schema Based Knowledge Graph for Clinical Knowledge Representation from Structured and Un-structured Oncology Data 1 Introduction 2 Background and Related Work 3 Design and Methodology 3.1 Data Analysis 3.2 KG Schema Construction 3.3 Information Extraction 3.4 Entity Selection 3.5 KG Construction 4 Results and Discussion 4.1 Name Entity Recognition 4.2 EHRs Visualization by Semantic Retrieval Approach 5 Conclusion and Future Work References Intelligent Fuzzy Clinical Decision Support System to Classify Breast Cancer—Case Study: The Wisconsin Dataset 1 Introduction 2 Material and Methods 2.1 Identifying the Dataset 2.2 Data Preparation (Crisp Inputs) 2.3 Reviewing Existing Models 2.4 Evaluating the Optimal Number of Clusters 2.5 Setting a Number of Clusters (Minimum and Maximum) According to the Previous Evaluation 2.6 Random Permutations 2.7 Cluster Analysis (Fuzzification Process) 2.8 Sampling Datasets (Cross-Validation or Random Sampling) 2.9 Pivot Tables 2.10 Elaborating the Decision Support System Based on Fuzzy Set Theory (Inference Engine) 2.11 Evaluating the Fuzzy System Performance (Defuzzification and Crisp Outputs) 3 Results and Discussion 4 Conclusions References Research on the Design and Production of VR Rehabilitation Game for Parkinson's Disease Patients Based on Real-Time Action Acquisition 1 Introduction 1.1 Rehabilitation Treatment of Parkinson's Disease 1.2 Rehabilitation Therapy with Kinect 1.3 Effects of VR Rehabilitation Games on Parkinson's Patients 1.4 VR Rehabilitation Games can be Played Remotely with the Help of the Network 2 Virtual Reality Rehabilitation Game Design 2.1 System Architecture of Butterfly Catching Game 2.2 Control the Number of Butterflies 2.3 Function Design for Simulating Butterfly Flying 2.4 Realization of Detecting and Catching Butterflies 2.5 Design of Database 3 Rehabilitation Game Making Based on Virtual Reality 3.1 Unity3D Components 3.2 Scene Production of Rehabilitation Games 3.3 Scene Production of Rehabilitation Games 4 Rehabilitation Game Making Based on Virtual Reality References Force-Directed Graph Layout Based on Community Discovery and Clustering Optimization 1 Introduction 2 Related Work 3 The Algorithm 3.1 Problems of Louvain Algorithm and the Optimization Idea Based on Pruning Idea 3.2 Implementation of Improved Louvain Algorithm 3.3 Problems of Force-Directed Graph Layout and the Clustering Optimization Idea 3.4 Group-Based Clustering Optimization Implementation 4 Experimental Results and Analysis 5 Conclusion and Future Work References Comprehensive Strategy to Screen the Ankylosing Spondylitis-Related Biomarkers in the Peripheral Serum 1 Introduction 2 Materials and Methods 2.1 Data Collection and Processing 2.2 DEGs Screening and GSEA Analysis 2.3 WGCNA-Based Targeting Modules and Genes Screening 2.4 Functional Enrichment Analysis 2.5 PPI Network Analysis 2.6 Multiple Algorithms Combine to Identify Prospective Biomarkers of AS 2.7 Validation of the Diagnosis-Related Gene Expression and Regulatory Mechanisms and Biological Functions of the Potential Biomarkers 3 Results 3.1 Identification of DEGs of Patients with Ankylosing Spondylitis 3.2 WGCNA-Based Filtering of Goal Modules and mRNAs 3.3 GO, KEGG and DO Enrichment Analyses 3.4 Establishing a PPI Network 3.5 Machine Learning Algorithm-Based Recognition of Prospective Biomarkers for AS 3.6 Validation of the Diagnosis-Related Gene Expression 3.7 Regulatory Mechanisms and Biological Functions of the Potential Biomarkers 4 Discussion References