The recent emergence of large language models (LLMs), which have unprecedented capabilities to process, analyze, and synthesize complex medical information with remarkable proficiency, is poised to have a disruptive impact on health care. In the field of medical imaging, LLMs can be applied with promise in generating radiology reports along with detecting and correcting errors, explaining medical imaging findings, indicating differential diagnoses based on imaging patterns, and providing recommendations on imaging modality and protocol selection. In parallel, LLMs could offer innovative solutions for individualized learning, intelligent tutoring, content generation, and clinical decision support in medical education. However, challenges such as incorrect responses, negative influence on critical thinking, academic integrity concerns, bias, and privacy issues must be addressed to ensure safe and effective implementation of LLMs. This review summarizes the current applications, potential benefits, inherent limitations along with appropriate mitigation strategies, and future directions of LLMs in medical imaging education, emphasizing the need for responsible integration to maximize their utilities while mitigating risks.
Zhu et al. (Wed,) studied this question.