Hemodynamic Data Assimilation in a Subject-specific Circle of Willis Geometry

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Fluid Dynamics

Background:

• The circle of Willis (CoW) anatomy varies significantly between individuals, leading to diverse intracranial vascularization and blood flow patterns.
• Accurate prediction of subject-specific hemodynamics is crucial for understanding cerebrovascular diseases.

Purpose of the Study:

• To develop and validate a data assimilation (DA) approach for predicting subject-specific hemodynamics in the CoW.
• To integrate 4D phase-contrast MRI (PC-MRI) data with computational fluid dynamics (CFD) simulations for enhanced accuracy.

Main Methods:

• The study employed a data assimilation technique, specifically the local ensemble transform Kalman filter (LETKF).
• This method was used to estimate the 3D transient velocity field within a subject-specific CoW geometry.
• This is the first known study to provide such a transient state estimate for the CoW using DA.

Main Results:

• Data assimilation reduced the uncertainty in velocity field estimates by up to 90%.
• The approach provided velocity data in areas below PC-MRI resolution, such as posterior communicating arteries.
• Uncertainty in wall shear stress distribution was reduced by a factor of two compared to CFD alone.

Conclusions:

• Data assimilation demonstrates significant potential for detailed vascular flow analysis.
• Combining diverse data sources using DA can statistically reduce uncertainty in hemodynamic estimates.
• This method enhances the understanding of blood flow dynamics in subject-specific CoW geometries.