Automated Optic Papilla Segmentation Approach Using Normalized Otsu Thresholding

In recent trends, "silent thief of sight" is the second leading cause of permanent vision impairment rivalling cataracts. This disease, Glaucoma, has no well-defined symptoms and hence it is difficult to diagnose at very early stage. In order to distinguish Glaucoma from a healthy eye, ophthalmologists usually compare the respective fundus images and try to identify the prolific growth of optic vessels in optic papilla. Usually, researchers use Hough transform to segment the optic disc from the fundus image that requires input from the external sources. In this paper, we have proposed an automated and optimized methodology with less computational effort that extracts the optic papilla without using Hough transform which aids in glaucoma detection. The proposed methodology involves the use of CLAHE, Normalized Otsu Thresholding, and a novel image extraction technique to obtain the optic disc from the fundus image. To demonstrate the effectiveness of this ideology, the algorithm is tested with High-Resolution fundus images dataset. On comparison, the experimental results reveal that the proposed methodology segments the optic papilla more accurately with reduced time complexity.