Efficient approximation of cardiac mechanics through reduced order modeling with deep learning-based operator approximation

Cicci, Ludovica; Fresca, Stefania; Manzoni, Andrea; Quarteroni, Alfio

doi:10.48550/arxiv.2202.03904

Cited by 2 publications

(1 citation statement)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Imaging techniques are combined with numerical simulations to perform robust parameter estimation in patient-specific cases [24,25,43,47]. On the other hand, multi-fidelity models of cardiac electromechanics, deep learning-based models of cardiac mechanics or simplified lumped circulation models are also employed for the same purpose [6,17,38,46]. All these mathematical tools mainly focus on the ventricular activity of the human heart.…”

Section: Introductionmentioning

confidence: 99%

Fast and robust parameter estimation with uncertainty quantification for the cardiac function

Salvador¹,

Regazzoni²,

Dedè³

et al. 2022

Preprint

View full text Add to dashboard Cite

Parameter estimation and uncertainty quantification are crucial in computational cardiology, as they enable the construction of digital twins that faithfully replicate the behavior of physical patients. Robust and efficient mathematical methods must be designed to fit many model parameters starting from a few, possibly non-invasive, noisy observations. Moreover, the effective clinical translation requires short execution times and a small amount of computational resources. In the framework of Bayesian statistics, we combine Maximum a Posteriori estimation and Hamiltonian Monte Carlo to find an approximation of model parameters and their posterior distributions. To reduce the computational effort, we employ an accurate Artificial Neural Network surrogate of 3D cardiac electromechanics model coupled with a 0D cardiocirculatory model. Fast simulations and minimal memory requirements are achieved by using matrix-free methods, automatic differentiation and automatic vectorization. Furthermore, we account for the surrogate modeling error and measurement error. We perform three different in silico test cases, ranging from the ventricular function to the entire cardiovascular system, involving whole-heart mechanics, arterial and venous circulation. The proposed method is robust when high levels of signal-to-noise ratio are present in the quantities of interest in combination with a random initialization of the model parameters in suitable intervals. As a matter of fact, by employing a single central processing unit on a standard laptop and a few hours of computations, we attain small relative errors for all model parameters and we estimate posterior distributions that contain the true values inside the 90% credibility regions. With these benefits, our approach meets the requirements for clinical exploitation, while being compliant with Green Computing practices.

show abstract

Section: Introductionmentioning

confidence: 99%

Fast and robust parameter estimation with uncertainty quantification for the cardiac function

Salvador¹,

Regazzoni²,

Dedè³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

A Deep Learning Model for the Identification of Active Contraction Properties of the Myocardium Using Limited Clinical Metrics

Nobrega

Mao

2023

Preprint

View full text Add to dashboard Cite

Technological breakthroughs have enhanced our understanding of myocardial mechanics and physiological responses to detect early disease indicators. Using constitutive models to represent myocardium structure is critical for understanding the intricacies of such complex tissues. Several models have been developed to depict both passive response and active contraction of myocardium, however they require careful adjustment of material parameters for patient-specific scenarios and substantial time and computing resources. Thus, most models are unsuitable for employment outside of research. Deep learning (DL) has sparked interest in data-driven computational modeling for complex system analysis. We developed a DL model for assessing and forecasting the behavior of an active contraction model of the left ventricular (LV) myocardium under a patient-specific clinical setting. Our original technique analyzes a context in which clinical measures are limited: as model input, just a handful of clinical parameters and a pressure-volume (PV) loop are required. This technique aims to bridge the gap between theoretical calculations and clinical applications by allowing doctors to use traditional metrics without administering additional data and processing resources. Our DL model's main objectives are to produce a waveform of active contraction property that properly portrays patient-specific data during a cardiac cycle and to estimate fiber angles at the endocardium and epicardium. Our model accurately represented the mechanical response of the LV myocardium for various PV curves, and it applies to both idealized and patient-specific geometries. Integrating artificial intelligence with constitutive-based models allows for the autonomous selection of hidden model parameters and facilitates their application in clinical settings.

show abstract

Efficient approximation of cardiac mechanics through reduced order modeling with deep learning-based operator approximation

Cited by 2 publications

References 55 publications

Fast and robust parameter estimation with uncertainty quantification for the cardiac function

Fast and robust parameter estimation with uncertainty quantification for the cardiac function

A Deep Learning Model for the Identification of Active Contraction Properties of the Myocardium Using Limited Clinical Metrics

Contact Info

Product

Resources

About