Foundation models for radiology: fundamentals, applications, opportunities, challenges, risks, and prospects

Abstract

Foundation models (FMs) represent a significant evolution in artificial intelligence (AI), impactin diverse fields. Within radiology, this evolution offers greater adaptability, multimodal integration and improved generalizability compared with traditional narrow AI. Utilizing large-scale pre-train ing and efficient fine-tuning, FMs can support diverse applications, including image interpretation report generation, integrative diagnostics combining imaging with clinical/laboratory data, an synthetic data creation, holding significant promise for advancements in precision medicine. How ever, clinical translation of FMs faces several substantial challenges. Key concerns include the in herent opacity of model decision-making processes, environmental and social sustainability issue risks to data privacy, complex ethical considerations, such as bias and fairness, and navigating th uncertainty of regulatory frameworks. Moreover, rigorous validation is essential to address inheren stochasticity and the risk of hallucination. This international collaborative effort provides a compre hensive overview of the fundamentals, applications, opportunities, challenges, and prospects o FMs, aiming to guide their responsible and effective adoption in radiology and healthcare.