Abstract
The optic disc is the starting point of optic nerves from the retina. It has a bright appearance in the retinal fundus image for a normal eye. In the case of some disease in the eye, optic nerves may get damaged or there can be other bright appearances in the retina whose intensity is similar or more than the optic disc. Optic disc detection is the preliminary step for most of the retinal disease diagnosis using computer vision. In the case of any eye disease, the task of optic disc detection becomes challenging. This paper proposes an automatic optic disc detection and segmentation technique to address the existing challenges. At first, the image is converted into a grayscale image and some preprocessing steps are applied to remove noise and normalize the contrast. An edge detection operation is performed to find the edges of the optic disc. Circular Hough transform technique is applied to find the possible optic disc center from the edges. Finally, a supervised machine learning algorithm is used to find the actual and perfect optic disc region among the candidate regions for the optic disc. The proposed method has been tested on 101 images from Drishti-GS retinal fundus image dataset. The overall system accuracy is 98.88% using the proposed approach.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have