Diabetic Retinopathy Detection Using Deep Learning

ABSTRACT If left untreated, diabetic retinopathy, a condition brought on by chronic diabetes, can cause complete blindness. Therefore, early detection of diabetic retinopathy is crucial, as is medical intervention to avoid its negative consequences. Ophthalmologist detection by hand takes longer and causes significant discomfort when being examined. One of the most well-liked methods for enhancing performance in a number of fields, including medical image analysis and classification, is machine learning. Consequently, an automated system helps identify diabetic retinopathy early. This study provides machine learning for the extraction of exudates, hemorrhages, and micro-aneurysms using a combination of neural networks. With an Accuracy of 85%,an artificial neural network model was able to identify diabetic retinopathy. Diabetic retinopathy has been successfully identified by a convolution neural network model, which also provides information on the disease's severity. The model's achieved Accuracy is 95%. KEYWORDS: Diabetic Retinopathy (DR), Machine Learning, Artificial Neural Networks and Deep Learning.