Vessel extraction is one of the critical tasks in clinical practice. Most of the deformable models used in these applications are based on either image gradient or global statistics of the intensity as their curve evolution and stopping criteria. However, the models using global statistic alone are not adaptive to the uneven intensity distribution since they do not take local variations in the area of interest into consideration. In this paper, we propose a novel level set based active contour model which combines a simple local function with the global statistical information for better vessel extraction results.