AI-Enhanced OCT Analysis for Detecting ILM Removal in ERM Surgery.

Purpose: This study evaluates how accurately humans and artificial intelligence (AI) can identify the type of surgery performed for epiretinal membrane (ERM) removal by analyzing postoperative optical coherence tomography (OCT) scans.

Methods: A retrospective analysis at the University of California San Diego included 250 eyes from 239 patients who underwent vitrectomy for idiopathic ERM between January 2013 and October 2024. Eyes were categorized into two groups: one with both the internal limiting membrane (ILM) and ERM removed using indocyanine green (ICG) staining, and another with only ERM removal, guided by triamcinolone. Postoperative OCT scans were labeled as either ERM-only or ILM+ERM peel based on surgical notes. Both the human grader and AI model were trained on 200 labeled OCT scans and tested on 50 masked OCT scans to classify the surgery type.

Results: Accuracy of the human grader in identifying the surgical technique was 50%, while the AI models demonstrated significantly higher accuracy. The ResNet18 model achieved 61±3%, while UwU-OrthLatt with DB4 initialization and UwU-PR-Relax with Symlet4 initialization reached 70±5% and 69±3%, respectively.

Conclusions: AI outperformed human grading in detecting ILM removal from OCT scans, demonstrating AI's potential in improving ophthalmic imaging for clinical use.