Classification of Stages Diabetic Retinopathy Using MobileNetV2 Model

Diabetic retinopathy (DR) is a complication of diabetes mellitus that causes retinal damage can lead to vision loss if not detected and treated promptly. The common diagnosis stages the disease take time, effort, cost be misdiagnosed. In recent period with explosion artificial intelligence, deep learning has become most popular tool high performance in many fields, especially analysis classification medical images. Convolutional Neural Network (CNN) more widely used as method imaging highly effective. this paper, five-stage image modern DR (healthy, mild, moderate, severe, proliferative) classified using technique. After cross-validation training testing on corresponding 5,590-image dataset, pre-MobileNetV2 model proposed classifying diabetic retinopathy. average accuracy achieved was 93.89% precision 94.00%, recall 92.00% f1-score 90.00%. thermal also given help experts for evaluating influence retina each different stage.