Detection of Diabetic Retinopathy Using Deep Learning-Based Framework

Abstract

Diabetic retinopathy (DR) is one of the common issues of diabetic mellitus that affects the eyesight of humans by causing lesions in their retinas. DR is mainly caused by the damage of blood vessels in the tissue of the retina, and it is one of the leading causes of visual impairment globally. It can even cause blindness if not detected in its early stages. To reduce the risk of eyesight loss, early detection and treatment are necessary. The manual process by ophthalmologists in the detection of DR requires much effort and time and is costly also. Many computer vision-based techniques reduce the manual effort for the automatic detection of DR. Machine learning is an important subset of computer vision mainly used in medical imaging for the detection of different diseases. This paper introduces a deep learning-based framework for the detection and classification of diabetic retinopathy, where we have trained a new customized convolutional neural network (CNN) model on two different benchmark datasets. This trained CNN model is tested on the separate test datasets. The detection performances of CNN are significantly encouraging, and it can assist to the doctors and radiologist in the early diagnosis of DR.KeywordsDiabetic retinopathyFundus imagesDeep learningConvolutional neural network