Validation and application of image-based CFD models of cerebral aneurysms

The increased availability of 3D in vivo aneurysm geometry data has led to computational fluid dynamics (CFD) modeling of patient-specific intracranial aneurysms. The primary goal of this thesis was to validate CFD predictions of these complicated flows. Initial validation compared in vivo digital subtraction angiography (DSA) against a virtual DSA reconstructed from the underlying CFD velocity field. Virtual angiographic images were found to be in excellent agreement with the corresponding clinical images, when the interaction between the injected contrast agent and local hemodynamics were properly modeled. To validate CFD more directly, we performed in vitro measurement of the intra-aneurysmal velocity fields using particle imaging velocimetry (PIV) of anatomically realistic flow-through phantoms. Before this could be done, we characterized representative input flow rate waveforms using phase contrast magnetic resonance imaging data previously acquired from normal subjects. The timing and amplitude of feature points from the individual waveforms were averaged together to produce an archetypal waveform shape appropriate for internal carotid (ICA) and vertebral (VA) arteries. PIV images were then collected on several planes, and CFD simulations were performed on micro-CT reconstructions of ICA and basilar tip aneurysm phantoms. PIV and CFD results for the ICA aneurysm showed good overall agreement. For the basilar tip aneurysm, both PIV and CFD similarly resolved the dynamics of counter-rotating vortices, as well as cycle-to-cycle fluctuations. A second patient-specific basilar tip model showed distinct intra-aneurysmal hemodynamics. We hypothesized that these distinct "hemodynamic phenotypes" could be anticipated by a simple geometric parameter: the angle made between the parent artery and the aneurysm bulb. We tested this hypothesis using an idealized basilar tip aneurysm model allowing independent control of the angle. Hemodynamics in the idealized model at 2 and 30 degrees were consistent with those in the corresponding patient-specific models, and the hemodynamic phenotype was found to switch between the angles of 8 and 12 degrees. The results of this thesis suggest that CFD can accurately model complex intra-aneurysmal flow dynamics. CFD may also be ideally suited for identifying simple-to-measure geometric factors, which may serve as clinically useful surrogate markers of specific hemodynamic phenotypes.;Keywords: Computational Fluid Dynamics, Intracranial Aneurysm, In vitro Validation, In vivo Validation, Digital Subtraction Angiography, Volumetric Flow Rate Waveform, Particle Image Velocimetry, Geometric Factors.