Removing ventricular far-field signals in intracardiac electrograms during stable atrial tachycardia using the periodic component analysis

Background: Intracardiac electrograms are an indispensable part during diagnosis of supraventricular arrhythmias, but atrial activity (AA) can be obscured by ventricular far-fields (VFF). Concepts based on statistical independence like principal component analysis (PCA) cannot be applied for VFF removal during atrial tachycardia with stable conduction. Methods: A database of realistic electrograms containing AA and VFF was generated. Both PCA and the new technique periodic component analysis (pi CA) were implemented, benchmarked, and applied to clinical data. Results: The concept of pi CA was successfully verified to retain compromised AA morphology, showing high correlation (cc = 0.98 +/- 0.01) for stable atrial cycle length (ACL). Performance of PCA failed during temporal coupling (cc = 0.03 +/- 0.08) but improved for increasing conduction variability (cc = 0.77 +/- 0.14). Stability of ACL was identified as a critical parameter for pi CA application. Analysis of clinical data confirmed these findings. Conclusion: pi CA is introduced as a powerful new technique for artifact removal in periodic signals. Its concept and performance were benchmarked against PCA using simulated data and demonstrated on measured electrograms. (C) 2015 Elsevier Inc. All rights reserved.