Deep Learning-Based Approach for Diabetic Retinopathy Detection with Explainable AI

الملخص

Diabetic Retinopathy (DR) is one of the most common complications of diabetes and a leading cause of vision loss among working-age adults worldwide. Therefore, early and accurate detection is crucial to preventing vision impairment and blindness. Automated deep learning-based diagnostic tools can play a transformative role in large-scale screening programs by enabling rapid, consistent, and cost-effective diagnosis of DR. This study explored the use of Convolutional Neural Networks (CNNs), specifically the VGG16 model, for detecting DR from retinal images and evaluating its diagnostic performance on a labeled dataset. Experimental results show that VGG16 performed strongly across all metrics, achieving an accuracy of 98.19%, precision of 98,21%, recall of 98.17%, and an F1-score of 98.18%, indicating robust and reliable performance in DR detection. In addition, this study applies Explainable AI (XAI) method—Occlusion —to improve the transparency and interpretability of deep learning models for DR detection. The findings highlight the importance of both accuracy and interpretability in building trust in automated diagnostics. By enabling early detection and supporting clinical workflows, the integration of high-performing models with XAI techniques offers a promising direction for reliable, AI-powered eye care solutions.