Active contour-based segmentation and removal of optic disk from retinal images

Abstract

A retinal fundus photograph is widely used in the diagnosis and treatment of various eye diseases such as diabetic retinopathy and glaucoma. Computer-aided analysis of fundus images provides an immediate detection and characterization of retinal features prior to inspection by a specialist. Segmentation of structures in such a retinal image can be used to detect and calculate the geometric shape and size of the optic disc and anterior segment with abnormal growth of any region in the eye. In this paper, we propose an algorithm based on active contours, often referred to as snakes, to remove the optic disk from retinal images. The proposed method consists of two main steps. In the first step, the optic disk boundary is approximated by means of edge detection, morphological operations and circular Hough transformation. In the second step, the exact boundary of the optic disk boundary is detected using an active contour model. The proposed algorithm was tested using 130 colored fundus images. Among those images, 20 were normal and 110 contained signs of the diabetic retinopathy. Results indicate a 90% accuracy in detecting the optic disc by the proposed technique.