ANALYSIS AND CLASSIFICATION OF RETINAL DISEASE SCREENING ON CIFAR-10 IMAGES USING CNN

Abstract: The rising prevalence of diabetes, driven by increased sugar consumption and machine-dependent lifestyles, has led to a significant rise in diabetic retinopathy (DR), a condition that often results in visual impairment. This project proposes an automated diagnosis system for DR by analyzing retinal (fundus) images using advanced image processing techniques. The system integrates a Convolutional Neural Network (CNN), along with AlexNet and ResNet architectures, to classify retinal images as either affected by DR or healthy. The developed graphical user interface (GUI) provides real-time predictions of DR severity and suggests appropriate actions, streamlining the diagnostic process and offering ophthalmologists a reliable decision support tool. Comparative analysis of different deep learning models is conducted to evaluate system performance, demonstrating its potential for enhanced diagnostic accuracy and reduced processing time in clinical applications. Keywords: Diabetic Retinopathy, Retinal Image, Image Processing, Convolutional Neural Network, Diagnostic process.