Computational fluid dynamics modeling of hemodynamics in the Circle of Willis during vasospasm and in the left ventricle in the presence of a left ventricular assist device

Abstract

Computational fluid dynamics (CFD) simulations can be leveraged to understand clinically relevant problems in cardiovascular flows. This dissertation explores two applications of CFD modeling to physiological flows: intracranial flow in the Circle of Willis (CoW) during vasospasm and intraventricular flow in the presence of a left ventricular assist device (LVAD). The CoW is a redundant network of blood vessels that perfuses the cerebral tissue. Flow in the collateral pathways that form this ring-like vascular structure can change in the presence of vessel constriction or occlusion. After a bleeding event in the subarachnoid space, vessels in the CoW sometimes involuntarily constrict, in a phenomenon known as vasospasm, which limits blood flow to the tissue, potentially causing infarct. The role of collateral pathways in the response to vasospasm is not well-understood. This dissertation investigates the relationship between changes in flow rate and direction in the collateral pathways, the anatomical variant of the CoW, and localization and severity of vasospasm across the network. Patient-specific CFD simulations were created in a cohort of 25 vasospasm patients, leveraging computed tomographic angiography (CTA) scans to generate models of the vasculature and transcranial Doppler ultrasound (TCD) measurements to apply boundary conditions. Bayesian analysis accounted for parameter uncertainty introduced by the medical data and was used to optimize the model parameters applied in the final simulation. Diameters, velocities, and flow rates were benchmarked against literature values, and virtual angiography performed by tracking a passive scalar, advected by the fluid velocity computed in the CFD simulations, was compared to clinical angiography, showing good agreement. Two metrics for vasospasm severity – percent changes in resistance and viscous dissipation – correlated closely with angiographic severity and helped identify regions of localization of vasospasm within the CoW and quantify overall severity. The second application of CFD in this thesis is on intraventricular flows. LVADs are centrifugal pumps implanted in the left ventricle (LV) of advanced heart failure patients. While pump designs have improved significantly over time, the risk of thromboembolic events remains high. Third generation pumps incorporate a pulsatility mode that modulates the rotational speed of the device impeller to promote in-pump washout. The role that this pulsatility mode plays in intraventricular washout is an active area of research. This dissertation studied how the temporal synchronization of the pulsatility mode with the native cardiac cycle affects the degree of intraventricular washout, which can have important implications for platelet activation and aggregation that subsequently can lead to thrombus formation. Lagrangian particle tracking was integrated into CFD simulations to model the hemodynamic environment experienced by platelets moving through the LV. Boundary conditions were defined using the time-varying flow rate from an equivalent particle image velocimetry (PIV) experiment. Regions of stasis were identified from examining velocity fields and calculating the stagnation index, which is an Eulerian quantification of stasis. Eulerian metrics from the CFD simulations agreed closely with PIV results. The optimal timing of the pulsatility mode with the cardiac cycle that maximizes intraventricular washout was identified.