Uncertainty quantification of a three-dimensional in-stent restenosis model with surrogate modelling

Area of Science:

• Cardiovascular Research
• Biomedical Engineering
• Computational Biology

Background:

• In-stent restenosis (ISR) is the re-narrowing of coronary arteries after stenting, often leading to angina or acute coronary syndromes.
• Vascular injury from balloon angioplasty and stent deployment triggers ISR, involving complex biological and mechanical processes.

Purpose of the Study:

• To quantify uncertainties in a computational model of in-stent restenosis.
• To identify key parameters influencing ISR progression and outcomes.

Main Methods:

• Developed a surrogate model using Gaussian process regression and proper orthogonal decomposition to handle high computational costs.
• Performed uncertainty quantification on a model with four uncertain parameters: endothelium regeneration time, smooth muscle cell bond breaking threshold, blood flow velocity, and internal elastic lamina fenestration percentage.
• Analyzed uncertainty propagation for two quantities of interest: average cross-sectional area and maximum relative area loss.

Main Results:

• Observed approximately 11% uncertainty in average cross-sectional area and 16% in maximum relative area loss.
• Higher internal elastic lamina fenestration was found to be a primary driver of uncertainty in neointimal growth during the early stages.
• Blood flow velocity and endothelium regeneration time emerged as major contributors to uncertainty in the later, clinically significant stages of restenosis.

Conclusions:

• Uncertainty quantification reveals critical parameters influencing in-stent restenosis.
• Early-stage ISR uncertainty is linked to vascular structure (fenestration), while later stages are influenced by dynamic factors (blood flow, regeneration).
• Findings guide future research and therapeutic interventions targeting specific stages of in-stent restenosis.