The use of artificial intelligence (AI) in radiology has rapidly gained attention in recent years. Radiology, being a digital image-based specialty, serves as the ideal testing ground for medical applications of AI. This, coupled with the increasing demand for clinical imaging and a shortage of radiologists, has led to a surge in AI-based presentations and articles in radiology conferences and journals and has necessitated the establishment of dedicated AI journals within the field. The majority of current research in AI in pancreatic imaging can be classified as either deep learning-based approaches or radiomics based approaches. Much of this novel research is already being integrated into clinical practice, with radiology leading all medical specialties in the number of federally approved AI tools with over 200 radiology-related AI programs approved by the Federal Food and Drug Administration as of the end of 2022 [1]. Accompanying the optimism surrounding this growth, however, are concerns about reproducibility, lack of generalizability and uncharted clinical translation of models reported in the literature. In addition, the relative infancy of AI, coupled with its “black-box” complexity, which demands multidisciplinary expertise in the peer review process, has led to a varied quality of results being published.

This purpose of this article was to review the current status of AI in pancreatic imaging and critically appraises the quality of existing evidence using the radiomics quality score.