Solving the Inverse Problem of Electrocardiography for Cardiac Digital Twins: A Survey

TL;DR

This paper aims to provide a comprehensive review of the methods for solving ECG inverse problems, their validation strategies, their clinical applications, and their future perspectives, including both conventional and deep learning-based techniques.

Abstract

Cardiac digital twins (CDTs) are personalized virtual representations used to understand complex cardiac mechanisms. A critical component of CDT development is solving the ECG inverse problem, which enables the reconstruction of cardiac sources and the estimation of patient-specific electrophysiology (EP) parameters from surface ECG data. Despite challenges from complex cardiac anatomy, noisy ECG data, and the ill-posed nature of the inverse problem, recent advances in computational methods have greatly improved the accuracy and efficiency of ECG inverse inference, strengthening the fidelity of CDTs. This paper aims to provide a comprehensive review of the methods of solving ECG inverse problem, the validation strategies, the clinical applications, and future perspectives. For the methodologies, we broadly classify state-of-the-art approaches into two categories: deterministic and probabilistic methods, including both conventional and deep learning-based techniques. Integrating physics laws with deep learning models holds promise, but challenges such as capturing dynamic electrophysiology accurately, accessing accurate domain knowledge, and quantifying prediction uncertainty persist. Integrating models into clinical workflows while ensuring interpretability and usability for healthcare professionals is essential. Overcoming these challenges will drive further research in CDTs.

Figures (9)

• Figure 1: Illustration of the electrocardiogram (ECG) inverse problem, including non-invasive cardiac source reconstruction, also known as ECG imaging (ECGI), and electrophysiological (EP) parameter estimation. Here, we take the biventricular modeling as an example. BSP: body surface potential; CV: conduction velocity.
• Figure 2: Summary of the ECG inverse inference methodologies. MCMC: Markov Chain Monte Carlo.
• Figure 3: Example of the physics-based model based on graph convolutional neural networks (CNNs) for the inverse inference of heart surface potential from body surface potential (BSP) journal/TMI/jiang2022. Illustrations designed referring to Jiang et al.journal/TMI/jiang2022 and Bear et al.journal/CAE/bear2018.
• Figure 4: Example of the MCMM application on the inverse inference of earliest activation sites (EASs) and CV from ECG. Here, MCMC has been used for the sampling of parameter sets of the population journal/MedIA/camps2021. Image adapted from Camps et al.journal/MedIA/camps2021 with permission.
• Figure 5: Illustration of Utah Torso Tank, with an isolated canine heart. Image adapted from Bergquist et al.journal/CBM/bergquist2021 with permission.