Solving the inverse problem in electrocardiography imaging for atrial fibrillation using various time-frequency...

Area of Science:

• Biomedical Engineering
• Computational Electrophysiology
• Medical Imaging

Background:

• Electrocardiographic imaging (ECGI) is crucial for identifying atrial fibrillation (AF) driving sources.
• Traditional ECGI inverse problem solutions lack multi-scale analysis and clinical reliability.
• Empirical Mode Decomposition (EMD)-based methods offer potential for time-frequency analysis in ECGI.

Purpose of the Study:

• To evaluate and compare various EMD-based solutions for the ECGI inverse problem in AF.
• To identify a more efficient and clinically reliable EMD-based approach for AF source localization.
• To assess the performance of UPEMD, IUPEMD, and EWT against other EMD methods using simulation and real patient data.

Main Methods:

• Applied five EMD-based solutions, including UPEMD, IUPEMD, and EWT, to AF simulation and clinical datasets.
• Evaluated performance using Pearson's correlation coefficient (CC), relative difference measurement star (RDMS), and distance (Dis) metrics.
• Assessed the accuracy of computed epicardial dominant frequency (DF) and driver probability (DP) maps, and source localization.

Main Results:

• UPEMD and IUPEMD demonstrated high CC (>0.95) and low RDMS (<0.3) for DF maps on simulation data.
• EWT also showed competitive performance for DF maps (CC >0.889, RDMS <0.48).
• UPEMD, IUPEMD, and EWT outperformed other EMD methods in DP map accuracy and source localization on real AF datasets, with UPEMD and EWT being superior.
• EWT exhibited the fastest signal deconstruction time (≤0.12s), followed by UPEMD (≤0.81s).

Conclusions:

• UPEMD-based and EWT-based solutions are superior for solving the ECGI inverse problem in AF.
• These methods demonstrate enhanced accuracy, efficiency, and reliability for clinical application.
• The study suggests UPEMD and EWT as promising tools for AF source identification in clinical settings.