Human illustrations surpass AI in accuracy for LASIK, PRK, and SMILE visuals
Although artificial intelligence (AI) platforms can generate visually appealing images, they fall short in creating accurate and detailed medical illustrations for refractive surgeries such as LASIK, PRK, and SMILE, according to a study. Human-created illustrations significantly outperform AI-generated images in terms of anatomical realism and procedural accuracy.
Researchers tested 2 AI image generators, DALL-E 3 and Medical Illustration Master, against human-created illustrations for accuracy. Six prompts were used to generate images, which were then graded by experts on factors such as anatomical realism and procedural accuracy.
Results showed that human-created illustrations consistently outperformed AI-generated images across all categories, including legibility and anatomical accuracy. For LASIK, PRK, and SMILE, human-created illustrations scored significantly higher than DALL-E 3 and Medical Illustration Master. Although AI platforms produced visually creative images, it lacked the precision required for medical education. The study concluded that AI, though promising in other fields, needs improvement in generating accurate medical illustrations.
Reference
Petroff DJ, Nasir AA, Moin KA, et al. Evaluating the Accuracy of Artificial Intelligence (AI)-Generated Illustrations for Laser-Assisted In Situ Keratomileusis (LASIK), Photorefractive Keratectomy (PRK), and Small Incision Lenticule Extraction (SMILE). Cureus. 2024;16(8):e67747. doi: 10.7759/cureus.67747. PMID: 39318903; PMCID: PMC11421855.