Improving prediction of the PVC origin non-invasively using the standard 12-lead ECG

Peter van Dam1, Ivo Strebel2, Roger Abächerli3
1UMC Utrecht, 2University Hospital Basel, 3HSLU


Background: RF-Catheter ablation is a procedure that is frequently performed in the case of premature ventricular contractions (PVCs). Preliminary localization of the PVC origin-site is critical to the precision-planning of the catheter route. A VCG-based method has proven to be as accurate in the prediction of the heart chamber as clinicians. The objective of the study was to improve the method such it will outer perform the clinicians.

Methods: A cohort of patients consecutively undergoing idiopathic PVC ablation was enrolled. The PVC origin-site was defined to be the site at which the procedure succeeded in eliminating the PVC. Patients with unsuccessful ablation or with scars were excluded. Every baseline 12-lead ECG was analyzed in a manually and blinded fashion by seven electrophysiologists (EPs) and the algorithm. Importantly, the VCG-based algorithms did not rely on patient-specific MRI models in order to have conditions equal to that of human experts.

Results: Of the 29 patients undergoing the analysis the median age was 55 years, of which 72% were females, median LVEF was 59% and median PVC-burden prior to ablation was 19%. The site of origin was localized to the right ventricle (RV) in 17 patients (59%). Manual analysis by the EPs achieved a mean accuracy of 81.3% (std: 5.8%, range: 72-90%) while the algorithm achieved an accuracy of 85.7% with an average error of 6.8mm (std: 5.2mm, range: 2.8-21.6mm).

Conclusion: The study results demonstrate that given VCG-based algorithm employing standard 12 lead ECG data and assuming a standard torso/heart model is capable of localizing the PVC’s site of origin to either heart chamber with an accuracy that is slightly higher than that of EPs (86% vs. 81%).