Artificial Intelligence in Breast Imaging: Challenges and Opportunities
Description
Artificial intelligence (AI) is transforming breast imaging by enhancing detection accuracy, improving workflow efficiency, and aiding in clinical decision support. Computer-aided detection (CAD) systems for mammography have been available since the late 1990s/early 2000s. These first-generation CAD tools were primarily used by radiologists to assist in the interpretation of mammograms, but modern AI systems are now designed to autonomously prioritize suspicious studies, reduce reading times, and risk stratification. Triage algorithms can flag high-probability cancer cases for expedited review or dismiss normal exams, thereby optimizing or reducing radiologist workload. Beyond mammography, AI is being integrated into tomosynthesis interpretation, ultrasound, and breast MRI, supporting lesion segmentation, quantitative analysis, and multimodal fusion. Despite promising results, challenges remain in generalizability, interpretability, and validation across diverse populations and imaging platforms. Regulatory pathways, such as the FDA’s Software as a Medical Device (SaMD) framework, continue to evolve to address continuous learning and performance monitoring. Overall, AI in breast imaging is transitioning from assistive CAD tools to dynamic, data-driven decision systems that aim to augment the expertise of radiologists and improve patient outcomes.