A Survey on Various Techniques for Detection of Diabetic Retinopathy in Retinal Fundus Images

Abstract: Diabetic Retinopathy refers to the retinal disease that occurs in Diabetic Mellitus Patients. It happens when the pancreas produces very little or no insulin. The blood vessels in retina will get blocked or grow without order in different directions. It can lead to vision loss over a period of time, if left untreated. Early detection of Diabetic Retinopathy (DR) is essential to treat DR effectively. The number of people suffering from Diabetics increased significantly over the last years. The traditional screening programs at an opticians Clinic is highly time consuming and the screening facilities available are less compared to the large population of Diabetic Patients in India( around 35 million).So the need for an effective computer based screening program arises, which can detect DR even in earlier stages. Various image processing techniques have been developed so far for the detection of DR. In this paper, a study of various DR detection methods and their comparisons are performed. The techniques can be broadly classified into optic disk segmentation, blood vessel detection, exudates detection and Microaneurysm (MA) detection methods. Different methods used are based on the feature to be extracted for DR detection. The image processing techniques are useful in identifying the location, size and severity grade of micro aneurysms and exudates in the retinal images. This survey gives an insight into the various prevailing methods for DR detection in retinal fundus images. Keywords: Diabetic Retinopathy, Microaneurysm, Image processing, retinal fundus images.