Vessel segmentation on angiogram using morphology driven deformable model

This paper presents an approach for the extraction of vessel on angiogram by using morphology feature driven deformable model. Segmentation of blood vessel on angiogram is a challenging task. The difficulty comes from both the low imaging quality and the complex shape of vessel. The low contrast within angiogram may make the edge detector based method produce false detection and cause the leakages problem of edge based deformable model, whereas the regions of interest in angiogram are not homogeneous, and therefore the popular region based deformable models, which usually take assumption of intensity homogeneity, are not applicable. To address these problems, we propose a vessel segmentation method based on region based active contour in which the morphology measure of regions acts as stop criterion of curve instead of gray intensity or its gradient and therefore the problem caused by gray intensity inhomogeneity is avoided. We present results on synthetic and real angiogram.