Comparison of dipole-based and potential-based ECGI methods for premature ventricular contraction beat localization...

Area of Science:

• Biomedical Engineering
• Computational Electrophysiology
• Medical Imaging

Background:

• Accurate localization of premature ventricular contraction (PVC) origin is crucial for guiding radiofrequency ablation (RFA).
• Non-invasive electrocardiographic imaging (ECGI) methods face challenges due to variations in source and forward models.
• The impact of torso inhomogeneities on PVC origin localization accuracy remains an area of investigation.

Purpose of the Study:

• To compare the performance of dipole-based and potential-based source models for localizing paced and spontaneous PVC origins.
• To evaluate the influence of torso inhomogeneities on the accuracy of these localization models.
• To assess the robustness of inverse methods used in ECGI for RFA guidance.

Main Methods:

• Utilized publicly available electrophysiology (EP) solution data (EDGAR) for paced PVCs and Bratislava data for spontaneous PVCs.
• Employed homogeneous and inhomogeneous torso models to simulate forward problems relating epicardial and endocardial sources.
• Evaluated localization error (LE) between true and estimated pacing sites/PVC origins.

Main Results:

• Dipole-based models showed median LE of 25.2 mm for paced and 30.2-33.0 mm for spontaneous PVCs.
• Potential-based models yielded median LE of 13.9 mm for paced and 28.9-39.2 mm for spontaneous PVCs.
• Torso inhomogeneities minimally affected dipole-based solutions but improved potential-based solutions on the ventricular surface.

Conclusions:

• The dipole-based source model demonstrates superior stability and robustness for PVC origin localization compared to the potential-based model.
• Torso inhomogeneities differentially affect the performance of PVC origin localization across different source models.
• Patient-specific geometric and forward models require careful consideration based on the chosen ECGI source model representation.