Enhancing Diabetic Retinopathy Screening through Advanced Retinal Image Analysis using Convolutional Neural Networks

Diabetic retinopathy (DR) is a prominent cause of vision loss in individuals of working age, making it crucial to identify the condition early for successful treatment. This paper presents an innovative approach using a convolutional neural network (CNN) model designed to automatically identify and diagnose diabetic retinopathy using high-resolution retinal pictures. The dataset consists of 2838 pictures classified into two classes: Diabetic Retinopathy and Not Diabetic Retinopathy. The data was split into three sets: training (2076 photos), validation (531 images), and test (231 images). The performance of the model was assessed by analyzing accuracy and loss graphs throughout the training and validation stages, as well as a confusion matrix during the testing phase. The confusion matrix was used to calculate key performance indicators like as Precision, Recall, F1 Score, and Accuracy. The CNN model attained an overall model accuracy of 94.37%, indicating its efficacy in differentiating between DR and non-DR pictures. The model's excellent accuracy demonstrates its promise as a dependable tool in scientific research for assisting healthcare professionals in the early diagnosis of Diabetic Retinopathy, enhancing patient outcomes, and alleviating the strain on healthcare systems. This work emphasizes the potential of deep learning models in medical imaging, which opens up opportunities for their wider use in automated illness identification and diagnosis.