Artificial Intelligence for Medical Imaging Technology: An Introduction

Foreword (For Euclid Seeram) Foreword (For Vijay Kanade) Preface Acknowledgments By Euclid Seeram By Vijay Kanade Contents Chapter 1: Artificial Intelligence in Medical Imaging Technology at a Glance 1.1 Introduction 1.2 A Little Bit of History 1.3 Definitions 1.3.1 Artificial Intelligence 1.3.2 Expansion of AI 1.3.3 Machine Learning 1.3.4 Deep Learning 1.4 Which Subfield of AI to Use in Medical Imaging? 1.5 Imaging Modalities Using AI Applications 1.6 How AI Works in a Nutshell 1.7 Applications of AI in Medical Imaging 1.8 AI-Based Image Reconstruction in Computed Tomography 1.9 Ethics of AI in Medical Imaging 1.10 The Radiological Technologist/Radiographer in AI 1.11 Deep Learning Implementation and Distinguishing Characteristics 1.12 Future Trends and Challenges References Chapter 2: AI Fundamentals 2.1 Introduction to Artificial Intelligence 2.2 Key Elements of AI 2.3 Definition of AI and Its Applications in Healthcare 2.3.1 Artificial Intelligence 2.3.2 Applications of AI in Healthcare 2.4 History and Evolution of AI 2.5 Overview of Machine Learning and Deep Learning 2.5.1 Machine Learning Overview 2.5.2 How ML Works? 2.5.3 Deep Learning Overview 2.5.4 How Deep Learning Works? 2.6 Clinical Decision Support Systems 2.6.1 Integration of AI Algorithms into Clinical Workflows 2.6.2 Challenges and Ethical Considerations in Implementing AI Systems in Healthcare 2.6.3 Interpreting and Validating AI Results for Clinical Decision-Making 2.7 Importance of AI in Medical Imaging References Chapter 3: Principles of Machine Learning 3.1 Introduction 3.2 Overview of Subfields 3.2.1 Supervised Learning 3.2.2 Unsupervised Learning 3.2.3 Reinforcement Learning 3.3 Key Components of a Machine Learning System 3.3.1 Data Acquisition and Preparation 3.3.1.1 Data Acquisition 3.3.1.2 Data Preparation 3.3.2 Model Architecture and Design 3.3.2.1 Model Selection 3.3.2.2 Model Architecture 3.3.2.3 Hyperparameter Tuning 3.3.2.4 Regularization and Optimization 3.3.2.5 Pretrained Models and Transfer Learning 3.3.2.6 Interpretability and Explainability 3.3.2.7 Validation and Evaluation 3.3.3 Loss Functions and Optimization Algorithms 3.3.3.1 Loss Functions 3.3.3.2 Optimization Algorithms 3.3.4 Training, Validation, and Testing 3.3.4.1 Training 3.3.4.2 Validation 3.3.4.3 Testing 3.3.5 Model Evaluation and Interpretability 3.3.5.1 Model Evaluation 3.3.5.2 Model Interpretability 3.3.6 Deployment and Continuous Monitoring 3.3.6.1 Deployment 3.3.6.2 Continuous Monitoring 3.4 Evaluation Metrics for Machine Learning Models References Chapter 4: Principles of Deep Learning 4.1 Introduction to Neural Networks and Deep Learning Architectures 4.2 Neural Network Fundamentals 4.2.1 Feedforward Neural Networks 4.2.2 Convolutional Neural Networks (CNNs) 4.2.3 Recurrent Neural Networks (RNNs) 4.2.4 Autoencoders and Unsupervised Learning 4.2.5 Generative Adversarial Networks (GANs) 4.2.6 Attention Mechanisms and Transformers 4.3 Convolutional Neural Networks (CNNs) for Image Classification 4.3.1 Motivation for CNNs in Image Classification 4.3.2 Convolutional Layers 4.3.3 Pooling Layers for Downsampling 4.3.4 Filter Sizes and Depth 4.3.5 Activation Maps and Receptive Fields 4.3.6 Transition to Fully Connected Layers 4.3.7 Dropout and Regularization 4.3.8 Transfer Learning with Pretrained CNNs 4.3.9 Data Augmentation Strategies 4.3.10 Interpretability of CNNs 4.4 Architectures for Segmentation, Object Detection, and Image Generation 4.4.1 Segmentation Architectures 4.4.1.1 U-Net Architecture 4.4.1.2 Fully Convolutional Networks (FCNs) 4.4.2 Object Detection Architectures 4.4.2.1 Region Proposal Networks (RPNs) 4.4.2.2 Single-Shot Detectors (SSDs) 4.4.3 Image Generation Architectures 4.4.3.1 Generative Adversarial Networks (GANs) 4.4.3.2 Variational Autoencoders (VAEs) References Chapter 5: Image Processing and Analysis 5.1 Introduction 5.2 Pre-processing Techniques 5.3 Segmentation Methods for Extracting Regions of Interest 5.3.1 Pixel-Level and Object-Level Segmentation Approaches 5.3.2 U-Net and Other Popular Architectures for Segmentation 5.4 Feature Extraction and Representation of Medical Images 5.5 Image Classification 5.5.1 Techniques for Categorizing Medical Images into Classes 5.5.2 Handling Imbalanced Datasets and Data Augmentation 5.6 Image Post-processing 5.6.1 Image Interpretation 5.6.2 Visualization and Presentation References Chapter 6: Artificial Intelligence Applications in Medical Imaging 6.1 Introduction 6.2 Radiology Workflow 6.2.1 ChatGPT and Radiology Workflow 6.3 Computer-Aided Detection/Diagnosis 6.4 Radiomics 6.4.1 Clinical Applications 6.5 Imaging Biobanks 6.6 Disease Detection and Classification 6.6.1 Framework for AI in Disease Detection Modeling 6.6.2 Evaluation of the Use of AI in Disease Diagnosis 6.7 Dose Optimization 6.7.1 AI in Fluoroscopy 6.8 Structured Reporting 6.8.1 Definition 6.9 Image Processing 6.9.1 Major Role of AI in Medical Image Manipulation and Comprehension References Chapter 7: Artificial Intelligence in Computed Tomography Image Reconstruction 7.1 Introduction 7.2 The Major Components of the CT Scanner 7.3 CT Image Reconstruction Fundamentals 7.3.1 The Filtered Back Projection Algorithm 7.3.2 Iterative Reconstruction Algorithm 7.4 Deep Learning Neural Networks: A Bare Bones Review 7.5 Deep Learning Image Reconstruction Algorithms 7.5.1 Basic Concept 7.5.2 CT Scanners Using DL Image Reconstruction Algorithms 7.6 Conclusion References Chapter 8: Computer-Aided Detection/Computer-Aided Diagnosis 8.1 Introduction 8.2 The Birth of CAD: CADe and CADx 8.3 The Main Stages of a CAD System 8.4 Artificial Intelligence and CAD 8.4.1 Deep Learning-Based CAD Systems References Chapter 9: Ethical and Regulatory Considerations 9.1 Introduction 9.2 Regulatory Frameworks and Guidelines for AI in Healthcare (e.g., FDA Regulations) 9.3 Privacy and Security Considerations for Medical Imaging Data 9.4 Bias, Fairness, and Transparency in AI Algorithms 9.5 Responsible Use of AI in Healthcare References Chapter 10: Future Trends and Challenges 10.1 Introduction 10.2 Recent Advances in AI for Medical Imaging 10.3 Emerging Technologies (e.g., Explainable AI, Federated Learning) 10.4 Potential Impact and Challenges in Adopting AI in Clinical Practice (e.g., Radiology) 10.5 Future AI: Considerations for AI Tool Development References Appendices Appendix 1 Appendix 2 Appendix 3 Appendix 4 References Index This book covers the principles, concepts, and applications of artificial intelligence in medical imaging technologies, specifically in the context of diagnostic imaging, such as radiography and radiological technology. First, artificial intelligence and its subsets machine learning and deep learning are described followed by a discussion of applications of these AI principles in medical imaging technologies. Finally, ethical questions, regulatory aspects, and future trends and challenges are also reviewed in this textbook. This book is intended for both students and practitioners in radiological technology, radiography, radiation therapy, nuclear medicine technology, diagnostic medical sonography, and biomedical engineering technology. Furthermore, residents in radiology, and medical physics students and related healthcare personnel (administrators and managers for example) may find this book useful.