Diabetic Retinopathy Detection using Machine Learning

Abstract

Diabetic Retinopathy (DR) is a leading cause of preventable blindness among diabetic patients worldwide [1]. Early detection and timely treatment are essential to reduce the risk of severe vision loss and improve patient outcomes. This research aims to develop an automated and efficient system for detecting and classifying DR using the YOLOv8 deep learning framework. The proposed methodology involves collecting retinal fundus images from publicly available datasets, followed by preprocessing techniques such as image resizing, normalization, noise reduction, and data augmentation to enhance image quality and improve model generalization. Transfer learning is utilized to extract meaningful features, and the YOLOv8 model is trained to classify images into multiple severity levels, including Normal, Mild, Moderate, Severe, and Proliferative DR.The experimental results demonstrate that the proposed model achieves a high accuracy of over 95%, along with strong precision, recall, and F1-score values. The system also supports real-time detection, making it suitable for clinical and large-scale screening applications. Furthermore, explainable AI techniques, such as heatmap visualization, are incorporated to highlight affected retinal regions, thereby improving model interpretability and clinical trust.In conclusion, this study highlights the effectiveness of deep learning in automating DR detection and emphasizes its potential to assist ophthalmologists in early diagnosis. Future work includes optimizing the model for mobile-based deployment and integrating it into real-time healthcare systems for broader accessibility.