Related Experiment Video
Updated: May 10, 2026
05:40
Using an Automated Hirschberg Test App to Evaluate Ocular Alignment
Published on: March 24, 2020
12.2K
Artificial Intelligence-Based Screening System for Diabetic Retinopathy in Primary Care.
Marc Baget-Bernaldiz1, Benilde Fontoba-Poveda2, Pedro Romero-Aroca1
1Ophthalmology Service, Hospital Universitari Sant Joan, Institut d'Investigació Sanitària Pere Virgili [IISPV], Universitat Rovira i Virgili, 43204 Reus, Spain.
Diagnostics (Basel, Switzerland)
|September 14, 2024
Summary
An artificial intelligence system accurately reads diabetic retinopathy in T2DM patients. A predictive algorithm effectively identifies patients at risk of developing diabetic retinopathy, aiding in early detection and management.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Diabetic retinopathy (DR) is a leading cause of vision loss in type 2 diabetic (T2DM) patients.
- Early detection and management of DR are crucial to prevent vision impairment.
Purpose of the Study:
- To evaluate an artificial intelligence-based reading system (AIRS) for classifying retinographies in T2DM patients.
- To assess a diabetic retinopathy predictive algorithm (DRPA) for predicting DR risk in T2DM patients.
Main Methods:
- AIRS was tested on 15,297 retinal images from a T2DM database and 1,200 from Messidor-2.
- DRPA was evaluated on 40,129 T2DM patients.
- AIRS and DRPA performance was compared to four retina specialists using sensitivity, specificity, accuracy, and AUC.
Main Results:
- AIRS achieved high accuracy (98.6%) in detecting referral DR (RDR) in the T2DM database and (96.78%) in Messidor-2.
- DRPA demonstrated strong performance in predicting the absence of DR (AUC = 0.92).
Conclusions:
- AIRS demonstrates excellent performance in reading and classifying retinographies for RDR in T2DM patients.
- DRPA shows effectiveness in predicting the absence of DR using clinical variables.
