Wall parameters in a carotid bifurcation phantom: comparison of 4D flow MRI measurements with computational fluid dynamics

Internal carotid artery (ICA) stenoses are a major source of stroke. Besides cardiovascular risk factors, plaque composition, and carotid bifurcation geometry, hemodynamic wall parameters (e.g., wall shear stress, WSS) are additional risk factors for the progression and rupture of ICA atheroma. 4D flow MRI allows measuring blood flow, however, only at limited spatial and temporal resolution. This hampers detailed and accurate calculation of wall parameters. The goal of this study is to investigate experimentally (using 4D flow MRI) and numerically (using two different CFD solvers) the hemodynamic parameters in a silicone phantom of a carotid bifurcation. By comparing significant parameters and flow patterns, the results can be cross-validated, and the quality of the 4D flow MRI data is assessed by comparison to data from direct numerical simulations. The phantom is perfused at a constant flow rate and the resulting flow field is captured by 7 Tesla MRI using 4D flow MRI using 4D flow MRI with an isotropic spatial resolution of 0.8 mm. For the CFD studies, the commercial CFD solver ANSYS Fluent with about 8.7 million cells and the open-source spectral element solver NEK5000 with 43,000 spectral elements are used. The flow patterns, averaged flow rate and wall shear stress magnitude obtained with these three methods have been compared. The average wall shear stress magnitudes in the bulb region are 0.001821 Pa (4D Flow MRI), 0.001203Pa (Fluent) and 0.001195Pa (NEK5000), respectively. The segmentation error, which leads to the inaccurate determination of the geometry and subsequently the flow parameters, can be one of the main factors of this difference. Despite these variations, the CFD and MRI results show similar WSS patterns. 4D flow MRI is capable of accurately determining flow patterns, but it falls short in quantifying secondary flows with lower velocities and recirculation regions in the bifurcation region. The results obtained from CFD are very consistent. This agreement between Fluent and NEK500 could be due to the laminar and steady nature of the imposed flow and lower quality results are expected from Fluent in a stenosed model with pulsatile flow where intermittent turbulence may occur. Accordingly, if the flow pattern becomes more complex, the quality of the MRI results can also decrease. Therefore, our ongoing work aims at combining 4D flow MRI measurements with CFD to enhance the quality of the predicted hemodynamic wall parameters