Creating a retinal image database to develop an automated screening tool for diabetic retinopathy in India

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Diabetic retinopathy (DR) is a leading cause of blindness globally.
• Annual DR screening is crucial for early detection and intervention in diabetic patients.
• Current DR screening methods face significant global challenges due to the large volume of patients requiring review.

Purpose of the Study:

• To address the need for comprehensive data and AI tools for DR screening.
• To establish a large-scale online Retinal Image Database in India for DR research.
• To develop a cost-effective, robust AI-based diagnostic tool for DR detection.

Main Methods:

• Initiated a multi-institutional study to create a large-scale online retinal image database in India.
• Collected high-resolution fundus images from various cameras, covering diverse DR lesions.
• Developed a framework for training AI algorithms using expert-annotated ground truths.

Main Results:

• Successfully initiated the creation of a substantial retinal image database.
• Established a methodology for developing an AI-driven DR diagnostic tool.
• Laid the groundwork for a scalable and efficient DR screening solution.

Conclusions:

• The developed retinal image database and AI framework are expected to significantly advance DR research and detection.
• This initiative promises a more efficient and scalable solution for managing the global challenge of diabetic retinopathy.
• The project aims to improve early DR detection and management through AI-powered diagnostics.