Lead PI:
Center Researchers:
The Challenge
Cardiovascular modeling and simulations require developing computational/mathematical models capturing the mechanistic behavior of arterial walls. While many constitutive models have been developed to represent the response of these tissues at various scales, the proper integration of inter- and intra-patient variability remains an open problem.
Neglecting response variability can lead to poor accuracy in predictions, with important consequences in terms of diagnostic and treatment.
To support the advancement of so-called digital twin frameworks, which typically involve a patient-specific replicate of the arterial geometry but do not account for intrinsic stochasticity in the system, there is a need to develop new computational methods and high-fidelity models capturing the variability in the mechanistic response of soft tissues.
Developing these models challenges the current the state of the art in computational mechanics, due to the variety of constraints that they must satisfy to ensure that predictions can be computed, as well as lack of results regarding the simulation of stochastic fields on complex domains.
Our Solution
The Guilleminot Lab has developed a series of high-fidelity probabilistic models that enable the integration of geometrical complexity and material stochasticity in near real time.
Leveraging recent results on numerical implicit representations for Gaussian fields, we provide a framework decoupling geometrical features and admissibility constraints for material parameters. The former are integrated on the fly, using ad hoc algorithms that can automatically process MRI inputs. The latter are introduced into an information-theoretical paradigm, which allows us to reduce bias in predictions.
We have shown, through thorough validation analyses, that the proposed framework can capture stochasticity in the response of human arterial layers, with and without spatial dependencies.
This work bridges the gap between the consideration of patient-specific geometrical models and the integration of material stochasticity, ultimately enhancing the accuracy and robustness of cardiovascular simulations in digital twin frameworks.
