Diabetic Retinopathy using Deep Learning

Diabetes is a metabolic disorder or more precisely Diabetes Mellitus (DM) happens because of the high level of blood sugar or say blood glucose in the body. Eye deficiency also is known as Diabetic Retinopathy (DR) is created by disease Diabetes which causes major vision loss. For successful treatment of Diabetic Retinopathy, it is very important to perform early detection which is one of the most essential challenges of its proper treatment. Even for trained clinicians or popular labs, the classification task of retinal images is found to be a tedious task. In many adjacent subjects, for diagnosis and treatment of diabetic retinopathy, a Convolutional neural network (CNN) have been used successfully. On the ImageNet Large Scale Visual Recognition Competition (ILSVRC), Deep convolutional networks have been achieving high-performance results on image classification challenge. Inception-V3 model is an example of a Deep Convolutional image recognition model. The unique feature of Inception-V3 is the extraction of different sized features of input images convolution level which is performed using Inception modules. We have used a pre-trained Inception-V3 model in this work because its Inception modules have many advantages for Diabetic Retinopathy Detection.