Artificial Intelligence System for Classification of Diabetic Retinopathy

Diabetic Retinopathy (DR) becomes the crucial disease in different disease groups and millions of people suffering with it every year rapidly. However, the conventional methods are failed to classify the DR in early stage due to complex architecture of eye fundus image. Therefore, this article is focused on implementation of deep learning convolutional neural network (DLCNN) based artificial intelligence approach for classifying multiple stages of DR. Initially, the hybrid features are extracted from IDRID dataset by using Local Binary Pattern (LBP), Local Gaussian Difference Extrema Pattern (LGDEP), and Histogram of Oriented Gradient (HOG) descriptors. Further, Linear Discriminant Analysis (LDA) is used to select the inter disease and intra disease dependent based optimal features. Then, DLCNN model is trained with these features for classification of DR grades for each test retinal image. The simulation results show that proposed DR classification results shows better subjective and object performance as compared to conventional machine learning and deep learning methods.