Investigating the use, role, and accuracy of generative artificial intelligence in anatomical illustration.

This study investigated the potential role of generative artificial intelligence (Gen AI) in creating and enhancing anatomical illustrations. Human made digital illustrations, Gen AI images created from text prompts, and Gen AI edited versions of the human made illustrations, were shared with staff and students at the University of Dundee with backgrounds in either medicine, dentistry, or anatomy. Participants were asked to view the illustrations and provide feedback regarding aesthetic preferences and accuracy, via an online survey. Twenty-five participants took part. Results were varied but demonstrated a small majority of participants preferred one of the Gen AI edited illustrations both overall and in terms of aesthetics and accuracy. In addition to the feedback survey, two anatomy lecturers were asked to rank each illustration out of five in terms of aesthetics, accuracy, and overall. The same Gen AI edited illustration came either joint first or second in all sections. However, in contrast to the survey, the human-made illustration was consistently ranked first or joint first. The difference in scores between the survey and anatomy lecturers suggests that the ability of the later to identify anatomical (in)accuracy influenced their decisions more strongly than stylistic preferences.