Machine learning in medical imaging

Abstract Medical imaging is an indispensable component of modern healthcare, playing a critical role in diagnosis, staging, and the assessment of treatment response for most major medical conditions. Advancements in image acquisition technologies have resulted in imaging modalities that capture more detailed and diverse visual information, leading to challenges in image analysis and interpretation. Within this context, machine learning (ML) algorithms offer the opportunity to enhance clinical analytics and decision-making through the creation of computer tools that can be trained to automatically analyze the visual details of medical image data. This chapter examines the current field of ML methods and their utility in clinical applications in decision support systems, with particular emphasis on advances from developments in deep learning. These methods are discussed with respect to the challenges in ML for medical imaging including limited annotated data, imbalanced class distributions, and labels that may be subjective or uncertain.