DIABETIC RETINOPATHY USING CNN RESNET-101

This abstract explores the application of Convolutional Neural Network (CNN) architecture, specifically ResNet-101, in the automated detection and classification of Diabetic Retinopathy (DR). Diabetic Retinopathy, a severe complication of diabetes, is a leading cause of vision impairment globally. Leveraging the deep learning capabilities of CNNs, particularly ResNet-101, this study aims to enhance the accuracy and efficiency of DR diagnosis through the analysis of retinal fundus images. The proposed model undergoes extensive training on diverse datasets to learn intricate patterns and features indicative of various DR stages. Through fine-tuning and transfer learning, the ResNet-101 architecture demonstrates its capacity to extract hierarchical features and nuances crucial for robust DR detection. Evaluation metrics, including sensitivity, specificity, and area under the receiver operating characteristic curve, assess the model's performance. The results showcase promising accuracy, demonstrating the potential of CNN-ResNet-101 as an effective tool for early and automated Diabetic Retinopathy diagnosis, offering a scalable and timely approach to addressing this critical healthcare challenge.

Key Words: diabetic retinopathy; image classification; deep convolutional neural network, ResNet-101.