Reduced-order modeling for cardiac electrophysiology. Application to parameter identification

TL;DR

This paper introduces a reduced-order POD-based model for cardiac electrophysiology that accurately approximates ECG features and is used for inverse parameter identification, including ionic parameters and infarction locations.

Contribution

It presents a novel reduced-order modeling approach for cardiac electrophysiology and demonstrates its effectiveness in inverse problems for parameter estimation.

Findings

Accurately approximates ECG restitution curves.

Efficiently identifies ionic parameters and infarction locations.

Reduces computational cost of cardiac simulations.

Abstract

A reduced-order model based on Proper Orthogonal Decomposition (POD) is proposed for the bidomain equations of cardiac electrophysiology. Its accuracy is assessed through electrocardiograms in various configurations, including myocardium infarctions and long-time simulations. We show in particular that a restitution curve can efficiently be approximated by this approach. The reduced-order model is then used in an inverse problem solved by an evolutionary algorithm. Some attempts are presented to identify ionic parameters and infarction locations from synthetic ECGs.