CUHK's AI-assisted diabetic macular edema detection system

The Chinese University of Hong Kong has developed an AI-assisted diabetic macular edema detection system. Clinical evaluation shows it improve the accuracy of screening and referrals, which could shorten patient waiting times.

Diabetic retinopathy, including diabetic macular edema, or DME, is a major cause of blindness among working-age individuals with diabetes.

If left untreated, it can cause irreversible blurry vision, and severe cases could lead to vision loss or blindness.

A research team from CUHK's Faculty of Medicine develops an AI-OCT system as an improved alternative to current screening standard for DME.

The system introduces new AI techniques to provide 3D imaging of the retina.

According to clinical evaluation, it shows high sensitivity and specificity in detecting DME.

It also improves the accuracy of screening and referrals.

60% of false positive cases can be screened out early using the system.

The team notes human intervention will remain part of the diagnosis to make sure no cases are missed, and they plan to furthur enhance clinical implementation.

CLEMENT THAM, Professor of CUHK Medicine: "There must be human in the loop. Because the AI system, at least for the time being, we cannot completely rely on AI system. So there must be an individual, an expert somewhere along the line. Basically we do not want to miss a genuine case of diabetic macular edema."

CAROL CHEUNG, Professor of CU Medicine: "We want to focus on real-time implementation, testing in the current workflow. What are the performance, and what are the concerns from different stakeholders."

Experts say the research offers practical framework for clinical specialties. It shows how AI-enabled tool can help in everyday patient care.