Automated detection of diabetic retinopathy using optimized convolutional neural network

Diabetes is one of the most common diseases across the world. It affects numerous parts of our body. Diabetic Retinopathy has an effect on retina which causes Diabetes Mellitus (DM) and it may even lead to blindness. Hence, detecting Diabetic Retinopathy (DR) is important during the early stages of diabetes which can prevent the patients from blindness. The publically accessible dataset of Diabetic Retinopathy (DR) contains numerous images of the retina and its results on Diabetic Retinopathy (DR). Our proposed ideology is to classify the images of the retina using an optimized convolutional neural network (OpCoNet) to detect whether the Diabetic Retinopathy (DR) is proliferative or severe or moderate or mild or normal. The optimized convolutional neural network has enhanced feature extraction and classification mechanism. Gray wolf optimization is used to train the CNN layers. The tested model is compared with the existing methodologies used for the detection of Diabetic Retinopathy (DR). The proposed technique effectually provides an accuracy of 98% and sensitivity of 98.5%. The automatic detection of Diabetic Retinopathy (DR) efficaciously proved in the screening process as well as lessens the trouble on medical services support.