Vessel Segmentation Using Shape Priors in Level Set Framework

In this paper,a new level set segmentation model is proposed,which is based on the shape priors that not need a large number of samples for training.The new level set segmentation model is aimed at vessel segmentation in a non-uniform image with weak object boundaries.First,we use the method of analyzing the mechanical stress tensor to analyse the Hessian matrix's features and define a vessel-ness function.Secondly,using the function to obtain the contour of shape,and level set signed distance implicitly expresses the shape curve.Finally,the minimal variance,FLUX and alignment for shape priors in energy function,and the extreme value of the energy function could be found by variational method.The curve evolution is dependent not only on the gradient and region information,but also on the anatomical constraints.The proposed method can locate the edge more accurate and accurately extract blood vessel.Experiments of vessel segmentation in images with intensity inhomogeneity confirm that the proposed algorithm yields more accurate segmentation results.