Text to image generators for anatomical illustrations: potential and limitations.

PURPOSE: In recent years, artificial intelligence has made significant advances in medical illustration, paralleling progress in many other fields. These tools provide an efficient and cost-effective approach to creating custom illustrations and rapidly visualizing medical and anatomical contents. The aim of this study is to investigate the strengths and limitations of artificial intelligence in generating such illustrations, with a focus on its potential application in anatomy education.

METHODS

The present study assessed the ability of ChatGPT, Microsoft Designer, Craiyon and Image FX to create illustrations of the liver, spleen, kidney, their cadaveric forms, and genital organs. Each illustration was independently evaluated by two researchers using a five-point Likert scale based on nine criteria assessing anatomical features, educational utility, and visual quality.

RESULTS

Image FX, ChatGPT, and Microsoft Designer achieved overall scores (3.46; 3.40; and 3.24, respectively), while Craiyon received the lowest score (1.86). Despite the aesthetic appeal, and visual quality of the generated images, a notable lack of anatomical features was observed. Microsoft Designer was restricted by policy from generating both cadaveric organ and genital system images, whereas Image FX, ChatGPT, and Craiyon produced such images despite their limited accuracy.

CONCLUSIONS

This study highlights both the capabilities and limitations of artificial intelligence powered text-to-image generators in anatomical illustrations. Future advancements in artificial intelligence models, particularly through the integration of anatomically accurate databases and standardized anatomical terminology, have the potential to significantly increase the accuracy and applicability of these tools in medical education.