Cerebrovascular occlusive disease (CVOD) is a major risk factor for ischemic stroke and is characterized by the presence of stenosis in the arteries supplying the brain. The cerebral vasculature has an innate ability to compensate for flow reductions, caused by the presence of CVOD, through a network of collateral pathways in the circle of Willis (CoW). However, flow compensation is highly patient-specific and dependent on the cerebral vasculature anatomy, availability of collateral pathways, degree of stenosis and, the condition of the cerebral microcirculation and its autoregulatory response. Unfortunately, clinically available imaging tools only provide limited information on flow compensation and the underlying cerebral hemodynamics. Given the complexity of the cerebral vasculature, better tools are necessary to characterize cerebral hemodynamics and guide the risk assessment of ischemic stroke.



Image-based computational fluid dynamics (CFD) provides a powerful tool for non-invasively analyzing cerebral hemodynamics with high spatial and temporal resolutions. However, CFD modeling of cerebral hemodynamics is challenging due to the need for patient-specific data to calibrate outflow boundary conditions in the brain. In this thesis, we explore a novel strategy to quantitatively characterize cerebral hemodynamics using CFD in combination with tissue perfusion from arterial spin labeling (ASL) MRI.



Firstly, we quantified territorial perfusion in the cerebral circulation through implementing and optimizing a vessel-selective arterial spin labeling (VS-ASL) sequence. VS-ASL is generally limited by its low labeling efficiency causing poor signal-to-noise ratio. We investigated the effects of off‐resonance, pulsatility, and vessel movement, and evaluated methods to maximize labeling efficiency and overall image quality. We found that an off-resonance calibration scan in combination with cardiac-triggering significantly improved labeling efficiency and image quality. Vessel movement during the MRI protocol occurred in the majority of study subjects and needs to be accounted for to maximize labeling efficiency.



Secondly, we developed a strategy to calibrate patient-specific CFD models of cerebral blood flow. The calibration consisted of estimating the total inflow to the CoW from PC-MRI and the flow splits in the CoW from non-selective ASL perfusion images. The outflow boundary conditions were iteratively tuned to match the estimated flow splits, and the ASL-calibrated CFD model was then validated against territorial perfusion maps from VS-ASL by calculating the blood supply to each cerebral territory using Lagrangian particle tracking (LPT). We found an overall good match in a small group of subjects; particularly, the flow compensation between hemispheres was captured well by the calibrated CFD models.



Thirdly, we investigated the impact of two outflow boundary condition strategies, an ASL-based and allometric-based calibration, on cerebral hemodynamics. The ASL-based calibrated CFD analysis captured the flow compensation between hemispheres as measured with VS-ASL and lead to an approximately symmetrical flow distribution in the CoW. In contrast, the allometric-based calibrated CFD analysis was unable to capture the collateral flow compensation, which resulted in large differences in flow between hemispheres.



Finally, the clinical feasibility and capabilities of our proposed CFD analysis was demonstrated in two CVOD patients. The CFD analysis showed significant differences in cerebral hemodynamics between the patients despite similar degrees of stenosis severity, highlighting the importance of a patient-specific assessment. Comparison of pre-operative and post-operative hemodynamics in one patient resulted in only minor changes following revascularization despite severe carotid stenosis. We demonstrated that our CFD analysis can provide detailed and quantitative information about hemodynamic impact of carotid stenosis and collateral flow compensation in the circle of Willis.


Date: Tuesday, May 4, 2021

Time: 3:00 PM

Zoom: https://umich.zoom.us/j/93059726229 (Zoom link requires prior registration)

Co-Chairs: Dr. C. Alberto Figueroa and Dr. Luis Hernandez-Garcia

For Assistance or Questions
um-bme@umich.edu