Deep-2DRP: Identifying Diabetic Retinopathy Progression via CNN-ResNet50

Abstract

Diabetic retinopathy (DR) is a dangerous condition that can result in vision loss for individuals with diabetes if it is not detected promptly. Identifying the specific stage of diabetic retinopathy is quite challenging and often necessitates the skilled interpretation of fundus images by professionals. Streamlining the revelation process is exigent and could typically benefit millions of individuals. The accustomed diagnosis method is based on experts manually scanning retinal pictures, which can be a rugged and error-prone procedure. This Study introduces Deep2DRP, a novel hybrid deep learning model that integrates CNN and Transfer learning with ResNet50 for exact DR prediction. The model demonstrated magnificent accuracy, with a success rate of 96.96%, precision of 0.97, recall of 0.96, F1 score of 0.97, and ROC- AUC score of 0.969. The outcome of this research has significant implications for ameliorating the diagnosis of DR. The Research highlights the dynamic of hybrid deep learning to significantly enhance Retinopathy prediction, providing valuable insight and tools for biomedical research and application. Future research will target ameliorating the elasticity of the model by incorporating a wider range of datasets and optimizing its integration with medical procedures to guarantee dependable and efficient performance in factual scenarios.