Optic disc detection via blood vessels origin using Morphological end point

Abstract

The location and boundary of optic disc is a fundamental step to analyze most of the retinal fundus images for finding the abnormalities occurred with the eyes. Most of retinal images have various sizes and shapes of optic disc due to the photographic nature and the sizes of patients' eyes. This leads to the difficulty in detecting the disc correctly. There are a number of ways to detect optic disc. In this work, we propose to extract optic disc boundary and its location using blood vessels origin. Due to the fact that optic disc will always be the origin of main blood vessels. This paper proposes a novel method for automatically detecting optic disc on retinal images. The proposed method consists of 4 main processes as follows 1) Improve image quality using image enhancement techniques on YIQ domain 2) Detect blood vessels using Tyler Coye algorithm and improve the result using Morphological together with feature extraction 3) Locate optic disc using Morphological end point 4) Draw the boundary of optic disc. The performance of the method is evaluated based on dataset from DIARETDB1 which is a public retinal image dataset. The performance accuracy is 91.11% compared with the ground truth.