| dc.description.abstract | Cardiovascular diseases have been the leading cause of death globally for the past decades. Many cardiovascular diseases are a result of structural changes in the arterial wall, which modify its stiffness. Thus, arterial stiffness has been established as a biomarker for cardiovascular heath. Current arterial stiffness measurements give insights into average vascular stiffness lacking a description of local properties. Adverse structural remodelling leads to significant local structural changes of the vessel wall. Understanding local vessel wall properties is therefore crucial for effective intervention and prevention strategies to mitigate cardiovascular events.
The overall project aims to develop a novel approach for assessing local arterial stiffness of the common carotid artery by combining non-invasive measurements with numerical modeling. This doctoral thesis focuses on developing a simple and cost-effective 1D-model which is validated against a 3D fluid-structure interaction model. Output uncertainty is quantified and a sensitivity analysis is conducted to explore model sensitivity with respect to uncertainties in model parameters.
Uncertainties are inherent in numerical models due to experimental limitations and unmeasurable parameters. Sensitivity analysis of a 1D-artery model with uncertainties in geometric and material parameters is conducted with polynomial chaos expansion revealing the Young's modulus and lumen radius as the most influential parameters on output uncertainty. High sensitivity of Young's modulus suggests potential for developing an inverse method using non-invasive measurements to dynamically estimate local arterial stiffness.
The number of model evaluations and computational expenses per model evaluation limit the use of established uncertainty quantification and sensitivity analysis methods to cheap numerical models. To address this, we investigated the feasibility of estimating Sobol' sensitivity indices using a multifidelity Monte Carlo method on an idealized artery model. Sobol' sensitivity measures were estimated by leveraging computational resources from a 3D-fluid structure interaction model to 1D- and 0D-representations. We validated the ability of the 1D and 0D-models to accurately predict pressure, flow rate, and radial displacement in comparison to predictions from the 3D-fluid-structure interaction model. Additionally, we showcased the utility of the multifidelity Monte Carlo method for conducting global sensitivity analysis with a 3D-fluid-structure interaction model serving as the highest-fidelity model.
Geometric and material parameters vary between age and sex-groups. To assess the 1D-model's applicability across sub-populations, we established through a structured literature review tailored input ranges for demographic groups, demonstrating sub-population-specific uncertainties. Uncertainty quantification and sensitivity analysis considered pressure, average diameter change, and pulse pressure as quantities of interest. Age affected pressure sensitivity, while sex influenced pulse pressure. Average diameter change showed consistent behavior, suggesting broad sub-population applicability of the 1D-model.
This doctoral thesis has made significant contributions to the advancement of a novel approach for assessing arterial stiffness through uncertainty quantification and sensitivity analysis using a 1D-model of the common carotid artery. It has further expanded the understanding of uncertainty quantification and sensitivity analysis methodologies for computationally expensive models. The findings suggest that integrating non-invasive measurements with a 1D-numerical model holds promise for estimating local arterial stiffness of the common carotid artery for broad sub-populations. This innovative approach could greatly impact cardiovascular disease prevention by enabling early detection of structural alterations in the arterial system, thereby providing clinicians with guidance for personalized treatment and intervention strategies. | en_US |
