Blood Vessel Segmentation Using Hybrid Median Filtering and Morphological Transformation

Abstract

Blood vessel segmentation in fundus images is an initial step for the identification of retina related diseases like diabetic retinopathy. The fundus images obtained from the patients show different parts of retina and abnormalities which depict disease presence. The work proposed in this paper presents an efficient vessel segmentation method using top-hat morphological transform and hybrid median filtering. The ability of the hybrid median filter to retain narrow lines and preserving corners is advantageous in the segmentation of the fine and twisted structure of blood vessels. The proposed method is tested on fundus images of publicly available databases and the performance of the proposed method is evaluated with respect to segmentation accuracy, specificity, and sensitivity and compared with other blood vessel segmentation methods in the literature.