A Statistical Shape Modeling Approach for the Derivation of a Data-Driven Geometry-Aware Lumped Arterial Stenosis...

Area of Science:

• Cardiovascular fluid dynamics
• Biomedical engineering
• Computational modeling

Background:

• Existing lumped arterial stenosis models lack accuracy for complex lesion shapes.
• This limits precise evaluation of coronary artery disease.

Purpose of the Study:

• To develop a geometry-informed, data-driven lumped stenosis model for accurate pressure-flow relationship prediction.
• To improve the estimation of trans-lesional pressure drops and fractional flow reserve (FFR).

Main Methods:

• Utilized statistical shape modeling (SSM) to create diverse synthetic coronary stenosis geometries.
• Employed high-fidelity 3D computational fluid dynamics (CFD) to derive reference pressure-flow data.
• Trained a lumped parameter model using CFD results and shape coefficients.

Main Results:

• A five-mode shape representation efficiently captured geometric variability.
• The new model significantly improved pressure drop prediction accuracy over conventional lumped models, especially for irregular morphologies.
• Integration into a 1D pulse wave propagation framework enhanced waveform correlation with CFD.
• Fractional flow reserve estimation improved by 18%.

Conclusions:

• The geometry-informed, data-driven lumped stenosis model offers superior accuracy for evaluating arterial lesions.
• This approach enhances the clinical utility of hemodynamic assessments, including FFR.
• The model's architecture supports future integration of patient-specific data for broader validation.