Improved Blood Vessel and Optic Disc Segmentation for Automatic Detection of Diabetic Retinopathy

A serious condition that threatens vision and results in blindness in persons of working age is diabetic retinopathy (DR). This work provides an inimitable approach for recognizing the optic disc (OD) established on the circular Hough Transform and mathematical morphology. The present work involves the performance of iterative blood vessel segmentation to segment the retinal veins and arteries for screening of DR. Retinal image processing requires both the optic disc and the vascular architecture, which act as landmarks for image identification. There is a lot of potential in the proposed method for automated retinal artery segmentation. This research implements segmentation in various phases that include blood vessels, optic disc, and exudates. Our method generates a two-dimensional odd-symmetric Gabor filter (GF). Finally, for DR, this work employs a Gabor-based texture descriptor in combination with Statistical Features, Colour Moments (CM), and Gray-Level Co-occurrence Matrix (GLCM).