Abstract: Diabetic retinopathy, also known as diabetic eye diseases which causes severe threat on sight. The detection of such abnormalities in the retina is known as diabetic retinopathy. It might reach to blindness in working age people. Many patients make diagnosis only after the vision is lost and vision loss is a late symptom of diabetic eye disease. By comparing retinal blood vessels with retinal vascular structures in retinal images, we can detect diabetes in early stages. In this paper, vessel segmentation is used to extract retinal image vessels. Morphological processing and modified kernel fuzzy c-means clustering have been used to segment the retinal blood vessels. Smoothing operation is performed to overwhelm the background information using mathematical morphology. After mathematical morphology, the enhanced image is segmented using modified kernel fuzzy c-means clustering algorithm. The proposed methodology is more effective than k- means clustering as it achieved accuracy 96.25%.

Keywords: Blood vessel extraction; Diabetic retinopathy; Modified kernel fuzzy c-means clustering; Mathematical morphology; Retinal image.