Optic Disc Segmentation in Fundus Images using Operator Splitting Approach

The optic disc (OD) is a critical anatomical pattern in retinal fundus images. The expertise to identify OD in fundus images play a significant part in automatic analysis frameworks. Any changes that occur in these anatomical structures result in retinal abnormalities. Hence it is essential to segment retinal features for detecting retinal diseases. Segmentation of the OD is significant for screening retinal disorder such as diabetic retinopathy(DR) and glaucoma. OD identification is a relatively easy task in standard retinal images, but Identifying this region in DR retinal images is difficult as Exudates-(EX) and OD appear the same features. In this paper, a novel technique using an operator splitting approach(OSA) for segmenting OD in fundus images with poor edges is proposed. A local database of 100 images collected from Silchar Medical College and Hospital(SMCH), Assam, and accuracy of 96% is obtained, which is better than other states of the art methods.