A particle image velocimetry (PIV) study relevant to verification and validation of a computational fluid dynamics (CFD) tool visualizing intracranial aneurysm hemodynamics
MetadataShow full item record
Computational fluid dynamics (CFD) is a common, non-invasive technique used for simulating and investigating the hemodynamic environment inside of intracranial aneurysms (IA). While various studies have theorized different roles of hemodynamics in vascular pathology and aneurysm treatment outcomes, the employed models have not been adequately validated for their intended use, hindering the realization of CFD as a routine clinical tool. Previous studies attempting to validate individual CFD models lacked implementation of pertinent verification and validation (V&V) techniques. These V&V methods stress the importance of quantitative validation metrics and uncertainty quantification throughout the process, which the previous IA validation studies have failed to address. The goal of this study was to introduce key concepts from past V&V studies and established V&V standards to support the validation of a computational model now used to visualize flow in IA. Recognizing that improved validation and verification are required for CFD modeling of complex blood vessel geometries, an experimental Particle Imaging Velocimetry (PIV) study was performed to identify circumstances where flow uncertainties might be excessively high. A patient-specific internal carotid artery (ICA) aneurysm geometry obtained from an actual clinical case was reproduced in a transparent silicone model, through which particle-seeded glycerol/water solutions were pulsed at velocities from 165 to 1000 mL/s and the PIV flow data was acquired for different pulse rates. Analysis of extensive experimental results illustrated the greatest flow variations to be at the highest flow rate, with lesser dependence on heart rates. Corner and edge features of the modeled aneurysm also were associated with greater flow uncertainties, as well as inflow/outflow zones of the aneurysm near the parent vessel. CFD simulations of these flow features can now be compared with the actual flow data as found by PIV measurements. Preliminary work along these lines has been initiated. Direct clinical input must determine the accuracy requirements for CFD, consistent with the V&V process.