Altered blood flow dynamics play a vital role in cerebrovascular diseases, such as cerebral small vessel disease (CSVD), an umbrella term encompassing various pathologies that affect small arteries, arterioles, capillaries, and venules. Understanding the complex relation between vascular geometry and local flow conditions requires high-resolution insight into microscale local hemodynamics. Nonetheless, in vivo imaging lacks the spatial resolution necessary to visualize flow in small vessels, and computational models require experimental validation to ensure modeling accuracy.

This thesis presents an integrated framework using 3D printed microfluidic models, particle image velocimetry (PIV), and computational fluid dynamics (CFD) to investigate hemodynamics on a microscale. Transparent 3D printed vascular models with simplified, straight, bifurcated, and pathological geometries were successfully fabricated using a direct 3D printing method employing a masked stereolithography (MSLA) 3D printer, achieving a minimum channel diameter of 0.5 mm. A developed post-processing method increased optical transparency, and a refractive index-matched working fluid was developed to minimize distortion. Microscopy and micro-CT imaging were used for morphological characterization, and flow experiments were conducted under steady-state laminar conditions. A microfluidic control system was employed to obtain global flow and pressure data, while micro-PIV was utilized to capture local velocity fields and wall shear stress (WSS). In the straight channels, the experimental results were compared with the analytical Hagen–Poiseuille approximation. The computational model was then validated using both analytical and experimental data for the straight geometries, followed by experimental validation of the computational model in the bifurcated and pathological geometries.

Velocity profiles between the computational and experimental results showed good agreement, exhibiting similar flow features with relative errors ranging from 5% to 17%. Experimental velocities were generally lower than those predicted by analytical and CFD methods, primarily due to limitations in near-wall resolution and visualization, as well as averaging over the depth-of-correlation (DOC) and the finite spatial resolution of PIV. These limitations were particularly pronounced in the sub-millimeter channel (0.5 mm) and at a lower magnification (4x objective). The wall shear stress comparison shows that the near-wall spatial resolution is the limiting factor rather than model physics. For the 4x objective or for the smaller-sized channels (0.5 mm and 1 mm channel), with a lower spatial resolution, the WSS was underestimated and showed significant deviations (15% to 40%) depending on the geometry. For the 2.0 mm channel and the middle section of the stenotic model at 10x magnification, the results matched well with an average error of 2.6%.

The thesis objective was achieved, and the integrated experimental-computational approach demonstrates that micro-PIV data from 3D printed vascular models can effectively be used to validate CFD simulations of microscale hemodynamics. The developed methodology provides a foundation for future work involving more complex patient-specific geometries, compliant walls, or non-Newtonian fluids, ultimately supporting the validation of numerical models for cerebrovascular flow under physiologically realistic conditions.