The application value of artificial intelligence in medical imaging: foundations, progress, and challenges

Abstract

Abstract Artificial intelligence (AI) is rapidly transforming medical imaging, reshaping the field as it does everyday life. As imaging data grow in complexity and clinical workloads continue to rise, AI has become increasingly essential for improving diagnostic efficiency and precision. However, despite these technical advances, the clinical adoption of AI in medical imaging remains fragmented, with limited generalizability, insufficient interpretability, and unclear clinical responsibility. In this review, we critically examine the evolution of AI in medical imaging, with a particular focus on its functional roles across the imaging workflow and its actual contributions to contemporary clinical needs. We systematically analyze current application paradigms, identify methodological and translational bottlenecks, and discuss emerging strategies to improve model robustness, interpretability, and clinical integration. In addition, we identify key barriers to the practical implementation of AI in medical imaging, including challenges related to cross-domain generalizability, data privacy issues and regulatory compliance. By examining these opportunities and challenges, this review aims to guide future research directions and foster deeper interdisciplinary collaboration among AI, clinical medicine, and biomedical engineering.