A Review on Recent Techniques For grading the Severity of Diabetic Retinopathy in Retinal Colour Fundus Images

Abstract

Diabetic retinopathy (DR) is an eye disease, which is caused by the development of retinal microvascularization following diabetes. It is a problem of diabetes mellitus, which produces lesions in the surface of the retina due to which eye vision gets affected. Severe, uncontrolled cases of diabetic retinopathy will result in blindness. Since DR cannot be reversed, it can lead to blindness, and only early treatment maintains vision. Early diagnosis and treatment of DR can significantly reduce The risk of losing the vision. Fundus images are manually examined for morphological changes in retinal lesions such as micro aneurysms, exudates, blood vessels, hemorrhages. They are a tedious and time-consuming job. It is often easily accomplished with the help of a computer-assisted system. The identification and classification of the severity of diabetic retinopathy requires adequate segmentation of the retinal lesions. In this article, various techniques for detecting retinal lesions are discussed for the final detection and classification of nonproliferative diabetic retinopathy. Blood vessel detection techniques for diagnosing proliferative diabetic retinopathy are also discussed. In addition, the available datasets for the fundus colored retina were also examined. This work will be useful for researchers and technicians who wish to use ongoing research in this area. Several challenging topics are also discussed that require further investigation.