Early Detection of Diabetic Retinopathy from Fundas Retinal Images using AI

Abstract

This study proposes an innovative method for detecting and grading diabetic retinopathy using publicly available fundus images. By integrating image processing and machine learning, it accurately identifies indicators like exudates and microaneurysms, crucial for assessing the condition's severity. Results show high accuracies, with support vector machine and K-Nearest neighbor methods achieving 92.1% for exudate grading, and decision tree model reaching 99.9% for microaneurysm grading. This automated approach holds promise for early detection and precise grading, facilitating timely intervention and improving patient outcomes.