Level set methods and MR image segmentation for geometric modeling in computational hemodynamics

Accurate, detailed simulation of important physiological blood flow features depends in part on accurate geometric models of the vascular structures to be studied. Because of their geometric nature, geometric image segmentation techniques provide powerful ways to improve the accuracy of geometric models in computational hemodynamics. Here we describe the application of one such technique, the level set method, to the construction of three-dimensional vascular models using MR imaging data. First, a boundary propagation function suitable for segmentation of vascular images is described. A methodology for applying this function in two-dimensions is outlined, and a strategy for using these two-dimensional segmentations to construct three-dimensional models is presented. Results of preliminary finite element blood flow simulations based on a computational model constructed in this way are also shown