AI-based DR severity screening tool shows comparable results to human reading centers

Purpose-built for physicians

LAS VEGAS — Ocula360, an AI-based tool for diabetic retinopathy lesion detection, showed efficacy in detecting DR across multiple levels of severity, according to a speaker at Clinical Trials at the Summit.

While several automated DR assessment tools have been FDA approved, there are limitations in screening for disease severity, SriniVas R. Sadda, MD, of Doheny Eye Institute, said.

“In the ideal world, we would actually automatically detect all of the lesions on these ultra-widefield images, and that could potentially facilitate a quantitative and continuous scoring system for diabetic retinopathy,” he said. “The good news is that there has been progress in this area.”

Sadda presented on Ocula360 (Ophthalytics), an automated algorithm trained on more than 487,000 images from four datasets, which uses a web-based platform to classify DR severity based on the seven standard ETDRS fields.