Optic Disc Segmentation from Colored Retinal Images Using Vessel Density

The accurate and precise segmentation of optic disc is a critical task in development of Computer Aided Diagnosis systems for analysis of colored fundus images to diagnose retinal diseases. The development of a reliable and efficient optic disc (OD) segmentation technique is a challenging task because of noise, illumination effects, blood vessels and lesions like exudates. In this paper, a novel technique is being presented for efficient and accurate segmentation of OD in the presence of noise and other lesions. This technique is comprised of four main modules, preprocessing, vessel segmentation, bright region extraction, and detection of OD. The system is verified and validated not only on publicly available databases of fundus image databases but also real images acquired locally. The validation of the proposed system was proved with the help of experimental results.