Vascular Tree Segmentation in Retinal Angiographies: Deformable Contour Model Approach

This paper describes a new method to detect the retina vascular tree. This detection is performed by a deformable contour model in angiographic retinal images. The generic deformable contour model is here redefined to obtain a more accurate segmentation by considering specific features of these vascular structures, mainly the anatomical landmarks that a crease extraction technique obtains. Results on sample images are promising, making it easier the computation of the AV index, very important for the diagnosis of several pathologies. Key–Words: Snakes, Retinal Fundus images, Segmentation, Tracing, Retinal Vasculature.