Generation of CFD meshes from biplane angiograms: an example of image-based mesh generation and simulation

Abstract

Physicians make extensive use of medical images for diagnosing and planning therapy, but they frequently confine themselves to a visual inspection of the morphology. However, medical images such as biplane angiograms could be exploited for clinically relevant simulation studies. We developed a simulation system that would allow physicians to assess a patient's hemodynamic state. This paper describes the automatic image-based generation of a mesh that is required for the patient-specific numerical simulation of the three-dimensional blood flow through a stenosed coronary artery using the finite element method. The first step in generating the mesh is the acquisition of the patient-specific geometry of the flow domain through a three-dimensional reconstruction based on biplane angiograms. We then aim at the generation of an optimal mesh that would allow us to compute the solution with a specified accuracy at minimal cost in terms of computing time. To do this, we must adapt the size of the elements to the flow conditions. As a consequent adaptive procedure with an a posteriori error analysis would consume too much time, we decided to employ a priori criteria for the adaptation. Although these criteria are in principle heuristic in nature, they nevertheless reflect a fair quantitative a priori knowledge relevant to the coronary artery under investigation. This quantitative knowledge is derived from a posteriori analyses of computed flow conditions in so-called reference flow domains.