Evaluation of magnetic resonance angiography as a possible alternative to rotational angiography or computed tomography angiography for assessing cerebrovascular computational fluid dynamics.

The aim of this study was to conduct a flow experiment using a cerebrovascular phantom and investigate whether magnetic resonance angiography (MRA) could replace three-dimensional rotational angiography (RA) and computed tomography angiography (CTA) to construct vascular models for computational fluid dynamics (CFD). We performed MRA and 3D cine phase-contrast (PC) MR imaging with a silicone cerebrovascular phantom of an internal carotid artery-posterior communicating artery aneurysm with blood-mimicking fluid, and controlled flow with a flowmeter. We also obtained RA and CTA data for the phantom. Four analysts constructed vascular models based on the three different modalities. These 12 constructed models used flow information based on 3D cine PC MR imaging for CFD. We compared RA-, CTA-, MRA-based CFD results using the micro-CT-based CFD result as the criterion standard to investigate whether MRA-based CFD was not inferior to RA- or CTA-based CFD. We also analyzed the inter-analyst variability. Wall shear stress (WSS) distributions and streamlines of RA- or MRA-based CFD and those of micro-CT-based CFD were similar, but the vascular models and WSS values were different. Accuracy in measurements of blood vessel diameter, cross-sectional maximum velocity, and spatially averaged WSS was the highest for RA-based CFD, followed by MRA-based and CTA-based CFD using micro-CT-based CFD result as the reference. Except maximum velocity from CTA, all other parameters had good inter-analyst agreement using different modalities. The results demonstrated that non-invasive MRA can be used for cerebrovascular CFD models with good inter-analyst agreements.