AI in Breast Imaging: Promise and Challenges

This chapter explores the current landscape of FDA-approved AI applications in breast imaging, emphasizing their potential to improve radiologist efficiency and accuracy. It presents a systematic review of 36 AI systems, revealing their limitations compared to individual radiologist accuracy, and discusses a Swedish trial showcasing AI's role in risk scoring mammography results. The chapter raises critical concerns about over-diagnosis and algorithm biases while highlighting the need for further research before AI integration into clinical practice.