Evaluation of multivariate adaptive non-parametric reduced-order model for solving the inverse electrocardiography...

Area of Science:

• Biomedical Engineering
• Computational Science
• Medical Physics

Background:

• The inverse electrocardiography (ECG) problem aims to determine cardiac electrical activity from body surface potentials.
• This is an ill-posed problem often addressed with regularization, optimization, or statistical methods.
• Spline-based techniques offer potential for complexity reduction and accuracy improvement in inverse problems.

Purpose of the Study:

• To evaluate the performance of Multivariate Adaptive Regression Splines (MARS) for solving the inverse ECG problem.
• To assess the MARS method's effectiveness in improving the accuracy and robustness of cardiac electrical activity reconstruction.
• To investigate the MARS method's capability in localizing arrhythmia sources.

Main Methods:

• Application of Multivariate Adaptive Regression Splines (MARS) to the inverse ECG problem.
• Utilizing two distinct collections of simulated data for method evaluation.
• Comparison of MARS-based solutions against existing approaches for inverse ECG.

Main Results:

• The MARS-based method demonstrated improved solutions for the inverse ECG problem.
• The MARS approach showed robustness against modeling errors.
• Significant improvements were observed in localizing arrhythmia sources using the MARS method.

Conclusions:

• Multivariate Adaptive Regression Splines (MARS) offer a promising non-parametric approach for the inverse ECG problem.
• The MARS method enhances the accuracy and robustness of reconstructing cardiac electrical activity.
• MARS is particularly effective for improving the localization of arrhythmia sources in clinical settings.