Optic disc and cup segmentation in fundus retinal images using feature detection and morphological techniques

Segmentation of optic disc and cup from the fundus retinal image helps in the detection of glaucoma which is one of the important causes for vision loss. In this study, we propose a method for optic disc followed by optical cup segmentation. The retinal input image is first preprocessed by applying histogram equalization and the connected components in the image are obtained. On applying the circle detection Hough transform, the circular-shaped optic disc is segmented from the other connected components. The morphological closing operation is applied on the segmented optic disc area to identify the optic cup location. The optic cup is then segmented by applying Otsu's thresholding only on the extracted green channel of the retinal image. The proposed method was evaluated using DRISHTI-GS dataset containing 50 test images. The performance metrics such as dice coefficient, average boundary localization error and cup-disc ratio error were obtained by comparing the proposed method with the ground truth values of 51 training images in the DRISHTI-GS dataset.