MR-driven Computational Fluid Dynamics

Introduction: Knowledge of blood flow patterns in the human body is a critical component in cardiovascular disease diagnosis. In-vivo blood flow is typically assessed either by 3D phase-contrast MRI [1], or by Computational Fluid Dynamics (CFD) calculations [2]. MRI measurements are direct, but offer limited spatio-temporal resolution. Conventional CFD calculations offer "infinite" spatial resolution, but rely on the accuracy of the assumed fluid properties and boundary conditions (e.g. blood viscosity, inlet/outlet flow, and the location and compliance of the vessel wall). We propose a hybrid MRI/CFD approach that integrates low-resolution MRI flow measurements directly into the CFD solver. The feasibility of the proposed “MR-driven CFD” approach is demonstrated in the carotid bifurcation of one healthy volunteer. We show that MR-driven CFD has a regularizing effect on the flow fields obtained with MRI alone, and produces flow patterns that are in better agreement with direct MRI measurements than CFD alone.