Innovative Automated Diabetic Retinopathy Severity Grading for Advanced Clinical Insights

Abstract

Diabetic retinopathy (DR) is a hazardous eye disorder that affects the retina and may lead to vision loss and blindness, especially in diabetics. Early identification is critical for a good outcome, however diabetic retinopathy can only be diagnosed through time-consuming and labor-intensive colour fundus pictures. In order to overcome this challenge, this study proposes a Deep Learning-based strategy that useIntelligent Diabetic Retinopathy Grading through Enhanced Neural Networks (IDR-ENN)to classify retinal pictures into distinct stages of diabetic retinopathy. The proposed approach was trained on a dataset that included 2200 photos from the testing set and 11000 coloured retinal images from the training set. The simulation results suggest that the IDR-ENNbased algorithm can achieve excellent levels of accuracy, sensitivity, and specificity.In this study, we propose a method to significantly reduce the computational time for diabetic retinopathy detection. A novelIDR-ENN approach achieves a remarkable 85% reduction in training computational time for diabetic retinopathy detection. The paper's overall conclusion underlines the potential of deep learning to improve the diagnosis and grading of diabetic retinopathy, which might have a significant impact on the prevention of blindness caused by this disease.