Non linear transformation field to build moving meshes for patient specific blood flow simulations

Our study propose a method to build patient specific geometric data and boundary conditions for unsteady CFD runs with variable meshes valid over the cardiac cycle. We develop a specific MRI protocol in order to extract the full geometric data at several phases over the cardiac cycle. Hemodynamic boundary conditions were acquired by means of another sequence into the same MRI device (velocity encoded MR imaging). Static vascular surfaces have been extracted by means of Level Set methods, after developing a noise reduction strategy called selective blurring filter. Finally, the mesh movement has been imposed to the static mesh according to dynamic MRI sequence by means of a non linear transformation field, computed from dynamic vascular images by means of bayesian algorithm. The proposed approach permits the computation of the blood flow under realistic in vivo, time evolving conditions. It is much simpler than the full coupled fluid structure problem and has the potential to provide a better picture of the specific hemodynamic status. Insights about the physiopathology of some arterial disease are also expected.