Autonomous Screening for Diabetic Macular Edema Using Deep Learning Processing of Retinal Images

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Diabetic macular edema (DME) is a leading cause of vision loss in diabetic patients.
• Early detection and treatment are crucial for preventing irreversible vision impairment.

Purpose of the Study:

• To develop and validate a deep learning model for detecting diabetic macular edema (DME) using color fundus imaging.
• To ensure the model's applicability in diverse clinical settings with various devices.

Main Methods:

• A deep learning model was trained on the large EyePACS dataset (over 32,000 images).
• Model performance was assessed using sensitivity, specificity, and AUC at image, eye, and patient levels.
• Independent validation was conducted on external datasets, including the Indian Diabetic Retinopathy Image Dataset.

Main Results:

• The model achieved high performance across all analysis levels.
• At the patient level, sensitivity was 0.900, specificity was 0.900, and AUC was 0.962.
• The model demonstrated robust performance on independent validation datasets.

Conclusions:

• Deep learning models can effectively detect diabetic macular edema from color fundus images.
• Automated DME detection can streamline screening processes for diabetic individuals.
• Further prospective studies are recommended to confirm clinical utility.