Method of detecting kink-bearing vessels in a retinal fundus image

Glaucoma is the second leading cause of blindness worldwide. The risk of glaucoma can be determined by calculating the cup to disc ratio in retinal fundus images. To accurately detect the optic cup, kinks or bends in small and medium vessels are important indicators of the cup boundary. In this paper, we present a method of detecting such vessels, through the extraction of patches and generation of hybrid features in a SVM-based model. The segmentation results show good potential for the further development of this method.