Uncertainty Quantification and Sensitivity Analysis for Computational FFR Estimation in Stable Coronary Artery...

Area of Science:

• Cardiovascular physiology
• Computational fluid dynamics
• Medical imaging analysis

Background:

• Fractional flow reserve (FFR) is crucial for diagnosing coronary artery disease.
• Accurate FFR estimation aids in treatment decisions.
• Reduced-order models offer potential for faster, patient-specific FFR predictions.

Purpose of the Study:

• To validate a reduced-order model for FFR estimation using patient-specific data and blood flow simulations.
• To assess the uncertainty in FFR predictions based on input data variability.
• To compare reduced-order model accuracy against full 3D simulations.

Main Methods:

• Utilized a hybrid 1D-0D reduced-order model for FFR prediction.
• Performed sensitivity analysis on model parameters using data from 13 patients.
• Compared reduced-order model results with 3D incompressible Navier-Stokes simulations.
• Characterized prediction uncertainty and identified influential input parameters.

Main Results:

• Satisfactory agreement between reduced-order model FFR and 3D model FFR was observed.
• Uncertainty in peripheral resistance reduction during hyperemia was the most influential factor for FFR predictions.
• Stenosis geometry uncertainty had a greater impact on FFR in low FFR cases.

Conclusions:

• Reduced-order model errors were smaller than input data uncertainties.
• Improving coronary blood flow measurements can significantly reduce FFR prediction uncertainty.
• This validated model provides a basis for more accurate computational FFR analysis.