A flexible framework for sequential estimation of model parameters in computational hemodynamics.

Area of Science:

• Computational fluid dynamics
• Biomedical engineering
• Cardiovascular modeling

Background:

• Patient-specific hemodynamic models are crucial for understanding cardiovascular diseases.
• Manual calibration of these models is time-consuming and limits clinical application.
• Accurate parameter estimation is essential for reliable hemodynamic simulations.

Purpose of the Study:

• To develop a flexible computational framework for efficient and automated parameter estimation in patient-specific hemodynamic models.
• To integrate data assimilation techniques for improved model calibration.
• To enhance the clinical applicability of 3D hemodynamic models.

Main Methods:

• Implementation of a Reduced-Order Unscented Kalman Filter (ROUKF) for wall material and lumped parameter network (LPN) boundary condition parameter estimation.
• Development of a constrained least squares augmentation (ROUKF-CLS) for complex LPNs.
• Utilizing a "Netlist" implementation for streamlined parameter filtering in intricate LPNs.

Main Results:

• Demonstration of the ROUKF algorithm using non-invasive patient-specific data (anatomy, flow, pressure) from a healthy volunteer.
• Validation of the ROUKF-CLS algorithm with synthetic data for a coronary LPN.
• Successful implementation of the developed methods within the CRIMSON hemodynamics software package.

Conclusions:

• The proposed computational framework significantly automates and enhances the parameter estimation process for patient-specific hemodynamic models.
• The ROUKF and ROUKF-CLS methods provide robust tools for data assimilation in cardiovascular flow modeling.
• This work facilitates more accurate and efficient construction of 3D hemodynamic models for clinical use.