Key Takeaways
- Adding AI-OCT to diabetic retinopathy screening reduced false-positive DME referrals from 69.1% to 24.1%.
- The AI-OCT system maintained 100% sensitivity for DME referral, matching standard screening.
- No cases of DME were identified among patients who were not referred in the AI-OCT group.
Adding an artificial intelligence (AI)–based optical coherence tomography (AI-OCT) system to diabetic retinopathy screening substantially reduced false-positive referrals for diabetic macular edema (DME) while maintaining referral sensitivity, according to a study.
The study included a prospective validation phase and a multicenter randomized clinical trial involving patients referred from a territory-wide diabetic retinopathy screening program.
In the validation cohort, the AI-OCT system achieved a sensitivity of 98.8% and a specificity of 90.7% for DME detection. A small proportion of scans were deemed ungradable (7.2%) or uncertain (4.4%).
In the randomized trial, 276 patients with suspected DME were assigned either to referral decisions based on both fundus photography screening and AI-OCT results or to standard referral based on fundus photography alone. Patients screened with AI-OCT had a false-positive referral rate of 24.1%, compared with 69.1% among those managed with standard screening alone.
Sensitivity for DME referral was 100% in both groups. Referral specificity was 86.5% in the AI-OCT group and 0% in the control group. Investigators reported that no cases of DME were identified among patients who were not referred in the AI-OCT group.
Reference
Zhang S, Ran A, Zhou J, et al. An AI-Based OCT System to Detect Diabetic Macular Edema: A Prospective Validation and Noninferiority Randomized Clinical Trial. JAMA. 2026 Jun 15:e267025. doi: 10.1001/jama.2026.7025. Epub ahead of print. PMID: 42295755; PMCID: PMC13270324.
Related Content
