AI Driven Diabetic Retinopathy Detection and Risk Assessment System Using Convolutional Neural Networks

Abstract

Diabetic retinopathy is a major healthcare concern that causes blindness, which makes it essential to detect the eye condition. The report discusses Retina Sense, a system to diagnose diabetic retinopathy and assess risk based on artificial intelligence techniques to create an Efficient Net B0 convolutional neural network and a clinical risk assessment module. The software was created with the integration of React Tan Stack for developing the front end, Node Js Express for the back end, and PostgreSQL database to automate retinopathy detection and risk assessment as well as treatment planning. The classifier was trained using the APTOS 2019 Blindness Detection dataset employing transfer learning and evaluated using its performance indicators.