Retinal vessels segmentation based on water flooding model

Abstract

Accurate segmentation of retinal blood vessels is an important task in computer aided diagnosis and surgery planning of retinopathy. In this paper, an unsupervised image segmentation of retinal vessels based on water flooding model is presented. The proposed approach imitates the nature of water flooding over land, where water always goes toward the low lands by the effect of gravity. The water flooding model supports water feeding to allow for covering more uncovered land regions and also allows for evaporation of water to help getting red of tiny regions or regions that may temporarily covered with water. The proposed vessel segmentation approach consists of three main phases. In the first phase, image image enhancement technique is employed to enhance the brightness corrected retina. Then a water flooding-based segmentation approach is applied to segment and extracts the retina vessel. Finally, a post processing phase is added to improve the results obtained from the segmentation phase using structural characteristics of the retinal vascular network. The proposed water flooding approach is tested on DRIVE databases of retinal images. The results demonstrate that the performance of the proposed approach is comparable with state of the art techniques in terms of accuracy, sensitivity and specificity.