Unsupervised Optic Cup and Optic Disk Segmentation for Glaucoma Detection by ICICA

Glaucoma is an eye disease that can lead to vision loss by damaging the optic nerve. Although this disease can often be prevented with early glaucoma detection, lack of discernible early symptoms makes the diagnosis difficult. Measuring the cup-to-disc ratio (CDR) is a common approach for glaucoma detection. Glaucoma can be specified by thinning the rim area that identifies the CDR value. Clustering and image segmentation can simply divide fundus images into distinct areas to estimate the optic disc (OD) and the optic cup (OC). This paper is based on a robust method, using the improved chaotic imperialistic competition algorithm (ICICA) for determining the position of the OD and OC on color fundus images for glaucoma detection. The predicted OD and OC boundaries are then used to estimate the CDR for glaucoma diagnosis. The performance of the proposed method was evaluated by using the publicly available RIGA dataset. It was found that some of the common problems of K-means clustering algorithm can be addressed by the proposed method for achieving better results. Moreover, the OC and OD regions can be precisely separated from the color image so that ophthalmologists can measure OC and OD areas more accurately.