Diabetic Net for Diabetic Retinopathy Image Classification Using Deep Convolutional Neural Network

Abstract

Among the most common diseases, diabetes is characterized by high blood sugar levels, which cause a slow human body erosion and affect vision. This disease affects the retina and can lead to blindness in some cases. There are a number of symptoms associated with this disease, such as enlarged blood vessels in the retina, bleeding, and aneurysms in small blood vessels. By detecting the disease early and taking measures to reduce its symptoms, it is possible to prevent these symptoms. For the detection of diseases, retinal images can be considered an indispensable tool. The wavelet transform was applied in this research to clarify diabetic retinopathy in order to improve the understanding of the retina. They were trained perfectly using computer vision and deep neural networks as compared to other neural network models. In order to classify retinoscopy images, a deep neural network (VGG16) was used. With SVM, KNN, Decision Tree, and Naive Bayesian methods, these papers significantly improved upon previous studies.