AJS: Federica Caforio

Speaker: Federica Caforio (University of Graz)
Title: Development and personalisation of multiphysics, multiscale models of cardiovascular function

Abstract:
Computational cardiovascular models offer a powerful approach for studying cardiovascular physiology and various pathological conditions. This field presents significant opportunities for research due to the complex and interdependent biophysical components underlying cardiac function. One key factor influencing the cardiac mechanical performance is its bidirectional interaction with the cardiovascular system. Existing 3D computational models of cardiac electro-mechanics (EM) often simplify the vascular system using 0D representations, which fail to capture important pulse wave transmission effects. Addressing this requires 1D models, but coupling 3D and 1D systems remains a challenge. In this presentation, we introduce a novel, stable approach for coupling a 3D cardiac EM model with a 1D blood flow model in the arterial system [1]. This represents the first development of a coupled 3D-1D model of the left ventricle and arterial system, demonstrated through numerical benchmarks to be robust and accurate across various time steps, and capable of simulating the physiological impact of arterial changes on pulse wave propagation. In the context of clinical application, we also highlight recent advancements in personalising this coupled model [2,3], a complex task given the available data and the computational costs of solving the inverse problem. Specifically, we present a new methodology that combines physics-informed neural networks [3] with advanced three-dimensional nonlinear cardiac biomechanical models to reconstruct displacement fields and estimate patient-specific biophysical properties, such as passive stiffness and active contractility.

A key feature of the proposed learning algorithm is that it solely relies on limited clinical datasets of displacement and, in some cases, strain data.Benchmark tests demonstrate the method’s accuracy, robustness, and its potential to efficiently estimate patient-specific biophysical parameters in nonlinear, time- dependent biomechanical models. This approach opens up new possibilities for detecting and characterising tissue inhomogeneities, such as fibrotic regions, and could drastically improve the diagnosis and treatment planning of cardiac conditions.

This work is in collaboration with Dr. Elias Karabelas, Matthias Höfler, Prof. Gundolf Haase from the University of Graz (AT); Dr. Francesco Regazzoni, Dr. Stefano Pagani, Prof. Alfio Quarteroni from Politecnico di Milano (IT); Dr. Christoph Augustin, Dr. Matthias Gsell, Prof. Gernot Plank from the Medical University of Graz (AT) and Dr. Jordi Alastruey from King’s College London (UK).

Скачать видео