Flow prediction in cerebral aneurysms based on geometry reconstruction from 3D rotational angiography

Abstract

We present an immersed boundary (IB) method for the simulation of steady blood flow inside a realistic cerebral aneurysm. We reconstruct a segment of the cerebrovascular system that contains an aneurysm, by using medical images obtained with three dimensional rotational angiography (3DRA). The main focus is on evaluating the sensitivity of flow predictions to the various steps of the vascular reconstruction process. Starting from the raw medical data, we analyze the fluid-mechanical consequences of the steps needed to generate the IB masking function for our simulations. We illustrate the IB method by applying it to a realistic aneurysm and investigate the role of (i) numerical resolution of the geometry; (ii) the selection of the specific vascular segment used in the simulations; and (iii) the influence of the smoothness of the periodic vessel extension to complete the computational model. Because of an unavoidable degree of uncertainty in the medical images, the geometry of the vessels and the aneurysm can be reconstructed only approximately. We also incorporate these slight uncertainties in the masking function by introducing inner and outer 'bounding' geometries and analyze the sensitivity of the flow predictions to these variations in the masking function. The numerical solutions computed in the inner and outer bounding geometries provide practical upper and lower bounds for basic flow properties, thus quantifying the reliability of the numerical solution, subject to uncertainties in the geometry of the flow domain.