Using Deep Learning to Identify Multiple Corneal Diseases from Eye Images

What this Trial Is About

Researchers are evaluating a deep learning algorithm designed to identify multiple corneal diseases using anterior segment images. The study aims to validate this algorithm by measuring its accuracy through sensitivity, specificity, positive predictive value, negative predictive value, and area under the curve. The study is observational and focuses on improving screening methods for corneal diseases. The intervention involves applying the artificial intelligence algorithm to diagnose corneal diseases from slit-lamp images. The algorithm analyzes key regions in the images, including the sclera, pupil, and lens, to detect disease presence. There are no treatment groups since this is a diagnostic study. Participants provide slit-lamp images that must meet quality standards, ensuring over 90% of the image area is clear for reading and discrimination. Researchers assess the algorithms performance by comparing its diagnoses with clinical standards within one week. Outcome measures include the area under the curve, sensitivity, and specificity. The study is conducted under the oversight of Tianjin Eye Hospital and does not involve healthy volunteers.

Artificial Intelligence for Screening of Multiple Corneal Diseases