Boundary Localization of Optic Disc using Pixel Level Deviation in Retinal Fundus Image

Abstract

In automated computer-aided diagnosis (CAD) approach for Diabetic retinopathy (DR) detection and monitoring, optic disc (OD) localization is the first and fundamental task for further processing of retinal fundus image. Identification of OD area is complex, as it attains the similar pathological blood vessels characteristics such as intensity variations and texture pattern. This paper presents a simple approach for the outer boundary localization of OD based on pixel level variations, where finding the centre of OD based on histogram equalization and work is carried out on MATLAB. This method significantly reduced the processing time and accuracy of OD segmentation process. We have tested two hundred images from DRIVE and STARE dataset (containing both normal and average diseased retina image), and notably attained the accuracy of 87%.

Keywords: computer-aided diagnosis (CAD), optic disc (OD), fundus image, diabetic retinopathy (DR), boundary localization

Cite this Article

Acharjya K, Agarwal S, Sharma SK. Boundary localization of optic disc using pixel level deviation in retinal fundus image. Research & Reviews: A Journal of Health Professions. 2018; 8(2): 14–18p.