Retinal Optic Disk segmentation and analysis in fundus images using DBSCAN clustering algorithm

The Optic Disk (OD) detection as an important part of the retinal images are considered for diagnosing the some diseases such as diabetic retinopathy, high blood pressure and hemorrhages also for exudes analyzing. In this paper, a novel method for optic disk detection and segmentation is proposed based on the modified clustering algorithm. In our method, at first step by applying the masking procedure on the image, the candidate of ROI for region of optic disk is obtained. This ROI is improved with the combinational filtering. The improved sub image for convenient presentation is converted to the LAB color space. The Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is used. In this clustering method with employing the square kernel with optimized parameters, the modified DBSCAN clustering is attained for the OD segmentation and recognition. Finally, for evaluating the proposed algorithm, several measures based on the structural content, similarity criterion and MSE of segmented images are computed. With this algorithm, the OD is finally segmented with the precision rate of 87.58% and the averaged variance of 0.0206 for the segmentation errors.