Vessel tree surface modeling for blood flow simulation

Angiography is a non-invasive medical examination of vessel system. The interpretation of resulting images is qualitative since the original data do not provide description of the vessels' geometry. To quantitatively diagnose disorders such as aneurysm or strictures one can use blood flow simulation. This requires high-quality three-dimensional models of vessels' geometry. We present a workflow consisting of algorithms for image enhancement, image segmentation, extraction of centerlines and estimation of diameters. It enables composition of three-dimensional vector models suitable for flow simulation. The approach was verified on contrast enhanced computed tomography images of kidneys and magnetic resonance time of flight images of brain.