Artificial Intelligence and Machine Learning for Healthcare: Vol. 1: Image and Data Analytics

Description

Artificial intelligence (AI) and machine learning (ML) have transformed many standard and conventional methods in undertaking health and well-being issues of humans.  AL/ML-based systems and tools play a critical role in this digital and big data era to address a variety of medical and healthcare problems, improving treatments and quality of care for patients.  This edition on AI and ML for healthcare consists of two volumes.  The first presents selected AI and ML studies on medical imaging and healthcare data analytics, while the second unveils emerging methodologies and trends in AI and ML for delivering better medical treatments and healthcare services in the future. In this first volume, progresses in AI and ML technologies for medical image, video, and signal processing as well as health information and data analytics are presented.  These selected studies offer readers theoretical and practical knowledge and ideas pertaining to recent advances in AI and ML for effective and efficient image and data analytics, leading to state-of-the-art AI and ML technologies for advancing the healthcare sector. Preface Contents 1 An Introduction to Artificial Intelligence in Healthcare 1.1 Introduction to Artificial Intelligence 1.2 Artificial Intelligence in Healthcare 1.2.1 Natural Language Processing (NLP) Technology 1.2.2 Machine Learning (ML) Algorithms 1.2.3 Artificial Neural Networks 1.2.4 Bayesian Classifier 1.2.5 Classification/Decision Trees. Random Forest 1.2.6 Survival Regression Models 1.2.7 Cluster Analysis 1.3 Advantages of Artificial Intelligence in Healthcare 1.4 Limitations of Artificial Intelligence in Healthcare 1.5 Successful Applications of Artificial Intelligence in Healthcare 1.6 Conclusions Appendix Books 2 Radiomics: Approach to Precision Medicine 2.1 Introduction 2.2 Materials and Methods 2.2.1 Building of a Database 2.2.2 Segmentation of Target Volume 2.2.3 Extraction and Selection of Useful Radiomics Features 2.2.4 Model Building Based on Machine Learning Technologies 2.3 Results and Discussion 2.4 Conclusions References 3 Artificial Intelligence Based Strategies for Data-Driven Radial MRI 3.1 Introduction 3.2 Related Work 3.2.1 Sparse Sampling Strategies 3.2.2 Contribution of the Manuscript 3.3 Problem Statement and Framework Description 3.3.1 Relationship Between Radial Projections and Image 3.3.2 Image Reconstruction, Resolution and Noise 3.3.3 Super-Resolution 3.3.4 Framework Details 3.3.5 Noise Threshold upper TT 3.4 Results and Discussion 3.5 Conclusion References 4 Unsupervised Domain Adaptation Approach for Liver Tumor Detection in Multi-phase CT Images 4.1 Introduction 4.1.1 Domain-Shift Problem 4.1.2 Domain Adaptation 4.2 Domain Adaptation Using Adversarial Learning 4.2.1 Anchor-free Detector 4.2.2 Proposed Multi-phase Domain Adaptation Framework Using Adversarial Domain Classification Loss 4.3 Proposed Multi-phase Domain Adaptation Framework Using Adversarial Learning with Maximum Square Loss 4.3.1 Maximum Square Loss 4.3.2 Overall Framework with Adversarial Domain Classification and Maximum Square Loss 4.4 Experiments 4.4.1 Implementation Details 4.4.2 Dataset 4.4.3 Evaluation 4.4.4 Results 4.5 Conclusions References 5 Multi-stage Synthetic Image Generation for the Semantic Segmentation of Medical Images 5.1 Introduction 5.2 Related Works 5.2.1 Synthetic Image Generation 5.2.2 Image-to-Image Translation 5.2.3 Retinal Image Synthesis and Segmentation 5.2.4 Chest X-ray Image Synthesis and Segmentation 5.3 Multi-stage Image Synthesis 5.3.1 Image Generation 5.4 Evaluation of Multi-stage Methods 5.4.1 Datasets 5.4.2 Segmentation Network 5.4.3 Experimental Setup 5.4.4 Two-Stage Method Evaluation 5.4.5 Three-Stage Method Evaluation 5.5 Conclusions References 6 Classification of Arrhythmia Signals Using Hybrid Convolutional Neural Network (CNN) Model 6.1 Introduction 6.2 Literature Review 6.3 Methodology 6.4 Results and Discussion 6.5 Conclusions Appendix 1 Appendix 2 Appendix 3 References 7 Polyp Segmentation with Deep Ensembles and Data Augmentation 7.1 Introduction 7.2 Related Methods 7.2.1 Overview of the Propose System 7.2.2 Loss Functions 7.3 Data Augmentation 7.3.1 Shadows 7.3.2 Contrast and Motion Blur 7.3.3 Color Mapping 7.4 Experimental Results 7.4.1 Data and Testing Protocol 7.4.2 Experiments 7.5 Conclusions References 8 Autistic Verbal Behavior Parameters 8.1 Introduction 8.2 Estate of the Art 8.3 Proposal, Materials and Methods 8.4 Testing Protocol 8.5 Analysis of Tests 8.6 Conclusions and Future Work References 9 Advances in Modelling Hospital Medical Wards 9.1 Introduction and Problem Addressed 9.2 Case Study and Data Analysis 9.3 Methodology and Results 9.4 Conclusion References 10 Tracking Person-Centred Care Experiences Alongside Other Success Measures in Hearing Rehabilitation 10.1 Person-Centred Care in Research and Practice 10.1.1 Situated Action—Understanding the Context as a Basis for Meaningful Measures 10.1.2 Situated AI for Achieving High-Quality Person-Centred Care 10.2 Co-design for Person-Centred Care Measures 10.2.1 Co-design of Evaluation Instruments 10.2.2 Artificial Intelligence and PCC 10.3 Case Study: Co-creation of PCC Measures and Dashboard with Hearing Rehabilitation Provider 10.3.1 Method 10.4 Results 10.4.1 Stakeholder Workshops—Development of Tools 10.4.2 Stakeholder Feedback 10.4.3 Piloting the Dashboard 10.5 Discussion 10.5.1 Summary of Case Study 10.5.2 Discussion on Opportunities and Challenges for AI 10.5.3 Quality of Data 10.6 Conclusions References 11 BioGNN: How Graph Neural Networks Can Solve Biological Problems 11.1 Overview of the Research Area 11.1.1 Biological Problems on Graphs 11.1.2 Deep Learning Models for Biological Graphs 11.2 Graph Neural Networks 11.2.1 The Graph Neural Network Model 11.2.2 Composite Graph Neural Networks 11.2.3 Layered Graph Neural Networks 11.2.4 Approximation Power of Graph Neural Networks 11.2.5 Software Implementation 11.3 Biological Applications 11.3.1 Prediction of Protein-Protein Interfaces 11.3.2 Drug Side-Effect Prediction 11.3.3 Molecular Graph Generation 11.4 Conclusions and Future Perspectives References