Automated localization of optic nerve head in a fundus image using computer vision techniques

Abstract

Glaucoma is an ocular disease which can be a cause of permanent blindness if not detected at an early stage. Computer aided diagnosis of diseases has been prevalent way of diagnosis in recent past. Correct localization of optic nerve head is important as detection of many diseases depends directly or indirectly on the correct detection of optic nerve head. The proposed method proposes a technique for correct localization of optic disc from fundus images. An anisotropic diffusion technique is used to smooth the images and extract pixels of sharp contrast like vessels and optic disc periphery. A statistical analysis is performed to highlight regions having high density of contrast pixels. Finally, morphological operations have been employed to determine the center of optic nerve head. The proposed has been tested on fundus images from DRIVE and MESSIDOR database. The applicability of the proposed method across images of different databases makes it suitable to be used in some real-time system in this field.