Artificial Intelligence in Medical Imaging - Opportunities, Applications and Risks

I've Seen the Future …

Preface

Contents

Part I Introduction

1 Introduction: Game Changers in Radiology

1.1 Era of Changes

1.2 Perspectives

1.3 Opportunities for the Future

1.4 Conclusion

Reference

Part II Technology: Getting Started

2 The Role of Medical Image Computing and Machine Learning in Healthcare

2.1 Introduction

2.2 Medical Image Analysis

2.2.1 Image Segmentation

2.2.2 Image Registration

2.2.3 Image Visualization

2.3 Challenges

2.3.1 Complexity of the Data

2.3.2 Complexity of the Objects of Interest

2.3.3 Complexity of the Validation

2.4 Medical Image Computing

2.5 Model-Based Image Analysis

2.5.1 Energy Minimization

2.5.2 Classification/Regression

2.6 Computational Strategies

2.6.1 Flexible Shape Fitting

2.6.2 Pixel Classification

2.7 Fundamental Issues

2.7.1 Explicit Versus Implicit Representation of Geometry

2.7.2 Global Versus Local Representations of Appearance

2.7.3 Deterministic Versus Statistical Models

2.7.4 Data Congruency Versus Model Fidelity

2.8 Conclusion

References

3 A Deeper Understanding of Deep Learning

3.1 Introduction

3.2 Computer-Aided Diagnosis, the Classical Approaches

3.3 Artificial Intelligence

3.4 Neural Networks

3.5 Convolutional Neural Networks

3.6 Why Now?

3.7 Example: Screening for Diabetic Retinopathy

3.8 Pointers on the Web

3.9 A Comparison with Brain Research

3.9.1 Brain Efficiency

3.9.2 Visual Learning

3.9.3 Foveated Vision

3.10 Conclusions and Recommendations

3.11 Take Home Messages

References

4 Deep Learning and Machine Learning in Imaging: Basic Principles

4.1 Introduction

4.2 Features and Classes

4.3 Neural Networks

4.4 Support Vector Machines

4.5 Decision Trees

4.6 Bayes Network

4.7 Deep Learning

4.7.1 Deep Learning Layers

4.7.2 Deep Learning Architectures

4.8 Conclusion

References

Part III Technology: Developing A.I. Applications

5 How to Develop Artificial Intelligence Applications

5.1 Introduction

5.2 Applications of AI in Radiology

5.3 Development of AI Applications in Radiology

5.4 Resources Framework

5.5 Conclusion

5.6 Summary/Take-Home Points

References

6 A Standardised Approach for Preparing Imaging Data for Machine Learning Tasks in Radiology

6.1 Data, Data Everywhere?

6.2 Not All Data Is Created Equal

6.3 The MIDaR Scale

6.3.1 MIDaR Level D

6.3.2 MIDaR Level C

6.3.3 MIDaR Level B

6.3.4 MIDaR Level A

6.4 Summary

6.5 Take Home Points

References

7 The Value of Structured Reporting for AI

7.1 Introduction

7.2 Conventional Radiological Reporting Versus Structured Reporting

7.3 Technical Implementations of Structured Reporting and IHE MRRT

7.4 Information Extraction Using Natural Language Processing

7.5 Information Extraction from Structured Reports

7.6 Integration of External Data into Structured Reports

7.7 Analytics and Clinical Decision Support

7.8 Outlook

References

8 Artificial Intelligence in Medicine: Validation and Study Design

8.1 The Validation of AI Technologies in Medicine

8.2 Safety in Medical AI

8.3 Assessing Model Efficacy Using Clinical Studies

8.3.1 The Clinical Question

8.3.2 The Ground Truth

8.3.3 The Target Population

8.3.4 The Cohort

8.3.5 Metrics

8.3.6 The Analysis

8.4 An Example of Study Design

8.5 Assessing Safety in Medical AI

8.6 Take-Home Points

References

Part IV Big Data in Medicine

9 Enterprise Imaging

9.1 Introduction

9.2 Basic Principles of Enterprise Imaging (EI)

9.3 Enterprise Imaging Platform

9.4 Standards and Technology for an Enterprise Imaging Platform and Image Sharing Across Enterprises

9.5 Legal Aspects

9.6 Enterprise Imaging in the Context of Artificial Intelligence

9.7 Take-Home Points

References

10 Imaging Biomarkers and Imaging Biobanks

10.1 Introduction

10.2 Stepwise Development

10.3 Validation

10.4 Imaging Biobanks

10.5 Conclusion

10.6 Take-Home Points

References

Part V Practical Use Cases of A.I. in Radiology

11 Applications of AI Beyond Image Interpretation

11.1 Imaging Appropriateness and Utilization

11.2 Patient Scheduling

11.3 Imaging Protocoling

11.4 Image Quality Improvement and Acquisition Time Reduction in MRI

11.5 Image Quality Improvement and Radiation Dose Reduction

11.6 Image Transformation

11.7 Image Quality Evaluation

11.8 Hanging Protocols

11.9 Reporting

11.10 Text Summarization and Report Translation

11.11 Speech Recognition

11.12 Follow-up

11.13 Worklist Optimization

11.14 Staffing Optimization

11.15 Business Intelligence and Business Analytics

11.16 Content-Based Image Retrieval

11.17 Patient Safety

11.18 Billing

11.19 Patient Experience

11.20 Challenges

11.21 Conclusion

11.22 Take-Home Points

References

12 Artificial Intelligence and Computer-Assisted Evaluation of Chest Pathology

12.1 Introduction

12.2 General Chest Radiography

12.3 Lung Nodules

12.3.1 Chest Radiography

12.3.2 Computed Tomography

12.4 Lung Cancer Radiomics

12.5 Pulmonary Embolism

12.6 Parenchymal Lung and Airways Diseases

12.7 Interstitial Lung Disease

12.8 Conclusions

12.9 Take-Home Points

References

13 Cardiovascular Diseases

13.1 Introduction

13.2 Impact of AI on Cardiovascular Imaging

13.2.1 Decision Support

13.2.2 Image Acquisition

13.2.3 Image Reconstruction and Improvement of Image Quality

13.2.4 Post-processing and Image Analysis

13.2.5 Interpretation and Diagnosis

13.2.6 Opportunistic Screening and Prognosis

13.2.7 Combining Imaging with Other Data Sources

13.3 Practical Use of AI in Different Cardiovascular Imaging Modalities

13.3.1 Echocardiography

13.3.2 Computed Tomography

13.3.3 Magnetic Resonance Imaging

13.3.4 Nuclear Imaging

13.3.5 Outcome Prediction Based on Composite Data

13.3.6 Deployment of Algorithms in Clinical Practice

13.3.7 Outlook and Conclusions

References

14 Deep Learning in Breast Cancer Screening

14.1 Background

14.1.1 The Breast Cancer Screening Global Landscape

14.1.2 The Rise and Fall of CAD

14.1.2.1 Rise: The Premise and Promise

14.1.2.2 How Does CAD Perform?

14.1.2.3 So Why Did CAD ``Fail''?

14.1.3 A Brief History of Deep Learning for Mammography

14.2 Goals for Automated Systems

14.2.1 Recall Decision Support

14.2.2 Lesion Localization

14.2.3 Density Stratification and Risk Prediction

14.3 Deep Learning Challenges Specific to Mammography

14.3.1 Memory Constraints and Image Size

14.3.2 Data Access and Quality

14.3.3 Data Issues During Training

14.3.3.1 Dataset Imbalance

14.3.3.2 Dataset Bias

14.3.3.3 Under-Fitting, Overfitting, and Generalization

14.3.4 Data Labeling

14.3.5 Principled Uncertainties

14.3.6 Interpretability

14.4 Future Directions

14.4.1 Generative Adversarial Networks (GANs)

14.4.2 Active Learning and Regulation

14.4.3 Tomosynthesis

14.4.4 Genomics

14.5 Summary

14.6 Take Home Points

References

15 Neurological Diseases

15.1 Introduction

15.2 Preprocessing of Brain Imaging

15.3 Applications

15.3.1 Protocoling, Acquisition, and Image Construction

15.3.2 Segmentation

15.3.3 Stroke

15.3.4 Tumor Classification

15.3.5 Disease Detection

15.4 Conclusion

15.5 Take-Home Points

References

16 The Role of AI in Clinical Trials

16.1 Introduction

16.2 Standardization of Medical Imaging in Clinical Trials

16.2.1 Before the Start of Clinical Trial

16.2.1.1 Image Acquisition Protocol Design

16.2.1.2 Site Validation

16.2.1.3 During the Clinical Trial

16.3 Artificial Intelligence in Clinical Trials

16.3.1 Classification Algorithms

16.3.1.1 Segmentation Algorithms

16.4 Digital Twin and In Silico Clinical Trials

16.5 Conclusion

16.6 Summary

References

Part VI Quality, Regulatory and Ethical Issues

17 Quality and Curation of Medical Images and Data

17.1 Introduction

17.2 Data Discovery and Retrieval

17.3 Data Quality

17.4 Adding Value

17.5 Reuse Over Time

17.6 Some Tools of the Trade

17.7 Conclusions

References

18 Does Future Society Need Legal Personhood for Robots and AI?

18.1 A Paradigm Shift

18.2 Legal Position

18.3 AI and Robots as Actor

18.4 Legal Subjectivity

18.5 Humans as (Natural) Legal Persons

18.5.1 Human-Like Behavior as Determination for Legal Personhood

18.5.2 Non-natural (Artificial) Legal Persons

18.6 Autonomous Artificial Intelligent Entities

18.6.1 AI in Robotic Entities

18.7 The Problem of Human–Robot Integration

18.8 An Alternative Personhood

18.8.1 Abstraction of the Legal Position of the Robot by a Narrative

18.8.1.1 The Cheshire Cat

18.8.1.2 The Reasonable Human

18.8.1.3 The Responsible Actor

18.8.1.4 Concluding on Legal Position

18.9 The Artificial Intelligent Entity or Robot as Legal Actor

18.9.1 Sui Generis Construct, Legal Subject or Legal Object Specialis?

18.9.2 Liability and Legal Subjectivity

18.9.3 Legal Acts

18.10 Where to Go from Here?

References

19 The Role of an Artificial Intelligence Ecosystem in Radiology

19.1 Defining Business Ecosystems

19.2 Artificial Intelligence Ecosystem for Healthcare and Diagnostic Imaging

19.3 Defining an Artificial Intelligence Ecosystem in Healthcare with a Focus on Diagnostic Imaging

19.3.1 Establish Realistic Goals

19.3.2 Maintain a Targeted Focus

19.3.3 Use High-Quality Datafor Training and Testing

19.3.4 Develop Consistent Methods for Validation and Monitoring Algorithm Performance

19.3.5 Build Public-Private Partnerships for Safety and Efficacy

19.3.6 Establish Standardsfor Interoperability and Pathways for Integration into Clinical Workflows

19.3.7 Promote Explicability of Algorithm Output

19.3.8 Facilitate Radiologist Input into Development, Validation, and Implementation

19.4 Bringing Artificial Products to Widespread Clinical Use: Challenges, Opportunities for Radiologists, and the Role of Medical Specialty Societies

19.4.1 Creating Clinically Effective Artificial Intelligence Use Cases

19.4.2 Enhancing the Availability of High-Quality Datasetsfor Algorithm Testing and Training

19.4.3 Maintaining Patient Data Privacy in Developing and Validating Artificial Intelligence Algorithms

19.4.4 Enhancing Algorithm Validation

19.4.5 Enhancing Clinical Integration

19.4.6 Mechanisms for Assessing Algorithm Performance in Clinical Practice

19.4.7 The Economics of AI and Business Models for Moving AI to Clinical Practice

19.4.8 Facilitating the Development of Non-interpretive Use Cases for Artificial Intelligence in Radiological Practice

19.4.9 Educating Non-radiologist Stakeholders About the Value of AI

19.5 Summary of the ProposedAI Ecosystemfor the Radiological Sciences

19.6 Conclusion

References

20 Advantages, Challenges, and Risks of Artificial Intelligencefor Radiologists

20.1 Innovation in Radiology

20.1.1 Artificial Intelligence (AI) Is the Next Big Thing

20.1.2 Radiologists' Perspective

20.2 Level of Expectation for AI in Radiology

20.2.1 AI Will Complement Many Routine Radiology Tasks

20.2.2 Will AI Also Surpass Existing Radiology Tasks?

20.3 Strategies to Prepare for the Future

20.3.1 Multitask Learning

20.3.2 Swiss Knife for Radiologists

20.3.3 Integration of Existing Medical Information Databases

20.3.4 Blockchain Technology

20.4 Hidden Risks and Dangers

20.4.1 Quality and Validation of Data

20.4.2 Data Security and Privacy

20.4.3 Ethics and AI

20.5 Take-Home Messages

References

AI: A Glossary of Terms

Glossary

Index