Man vs machine: Performance of manual vs automated electrocardiogram analysis for predicting the chamber of origin of idiopathic ventricular arrhythmia. (25th December 2019)

• Record Type: Journal Article
• Authors: Asatryan, Babken; Ebrahimi, Ramin; Strebel, Ivo; van Dam, Peter M.; Kühne, Michael; Knecht, Sven; Spies, Florian; Abächerli, Roger; Badertscher, Patrick; Kozhuharov, Nikola; Zeljkovic, Ivan; Schaer, Beat; Osswald, Stefan; Sticherling, Christian; Reichlin, Tobias
• Is Part Of: Journal of cardiovascular electrophysiology. Volume 31:Number 2(2020)
• Page Start: 410
• Page End: 416
• Publication Date: 2019-12-25
• Subjects: catheter ablation -- computational prediction -- ECG -- idiopathic ventricular arrhythmia -- premature ventricular complex; Blood vessels -- Physiology -- Periodicals; Electrophysiology -- Periodicals; Heart -- Physiology -- Periodicals; 612.1
• DOI: 10.1111/jce.14320
• Languages: English
• ISSNs: 1045-3873

Abstract

Abstract: Background: Radiofrequency catheter ablation of idiopathic ventricular arrhythmias (VAs) is performed to eliminate symptoms and to prevent or reverse arrhythmia‐induced cardiomyopathy. Preprocedural prediction of the chamber of VA origin is critical for patient counseling, procedure planning, and guidance of invasive mapping. Objective: We aimed to assess the performance of manual expert versus automated 12‐lead electrocardiogram (ECG) analysis in the prediction of VA origin. Methods: Patients with ablation of idiopathic VA and sustained success were included. The VA origin was defined as the site where ablation caused arrhythmia suppression. Standard baseline 12‐lead ECGs with documentation of the VA were analyzed manually in a blinded fashion by three electrophysiologists and three electrophysiology (EP) fellows. In addition, the same standard 12‐lead ECG was analyzed by an automated computer algorithm using a vectorcardiographic approach. Results: Thirty‐eight patients (median age, 47 [interquartile range, 37–58]; 68% female) were enrolled. The VA originated from the right ventricle in 24 (63%) and the left ventricle in 14 (37%) patients. The electrophysiologists and EP fellows identified the VA chamber of origin with a similar accuracy of 73% and 72% ( P = .72). The automated algorithm showed a higher accuracy of 89% ( P = .03 compared with electrophysiologists and EP fellows). This resulted in a sensitivity of 95% and specificity of 86%. Conclusion: While the manual ECG analysis of the standard 12‐lead ECG by both electrophysiologists and EP fellows correctly identified the chamber of VA origin in around 75% of cases, an automated vectorcardiographic computer algorithm achieved an accuracy of 89% with clinically acceptable diagnostic parameters.