Introduction: Atrial activity on the surface ECG during premature beats and supraventricular arrhythmias frequently is obscured by the superimposed QRST complex of the previous cardiac cycle. This study examines the performance of a newly developed automatic QRST subtraction algorithm to isolate ectopic P waves from the preceding T-U wave.
Methods and results: The 62-lead ECG recordings were obtained during (1) sinus rhythm and programmed right atrial stimulation in 12 patients (group A); and (2) sinus rhythm and atrial premature beats, atrial tachycardia, or paroxysmal atrial fibrillation in 5 patients (group B). Pacing in group A patients was conducted at a slow drive cycle length to generate an ectopic P wave not obscured by the previous QRST complex and by delivering single premature extrastimuli at progressively shorter coupling intervals to produce an ectopic P wave obscured by the upsloping (early T-U wave), peak (middle T-U wave), and downsloping component of the T-U wave (late T-U wave). All ectopic P waves in group B patients were concealed by the preceding T-U wave. Automatic QRST subtraction was attained using an adaptive template constructed from averaged QRST complexes (mean 83 +/- 25 complexes) obtained during sinus rhythm (groups A and B) or atrial overdrive pacing (group A). P wave integral maps subsequently were computed, visually compared, and mathematically correlated. A high correspondence in spatial map pattern was observed between integral maps of "nonobscured" and previously "obscured" paced P waves obtained in group A patients (mean r = 0.88 +/- 0.07) as well as between integral maps of two to three previously obscured P waves with the same atrial arrhythmia morphology obtained in group B patients (mean r = 0.94 +/- 0.05). Improved morphologic P wave replication in group A patients was acquired when concealment occurred in the early (mean r = 0.90 +/- 0.08) or late part of the T-U wave (mean r = 0.90 +/- 0.06) as opposed to the middle T-U wave (mean r = 0.85 +/- 0.07) (P = NS and P < 0.05 for early vs middle and late vs middle T-U wave, respectively).
Conclusion: This novel automatic 62-lead QRST subtraction algorithm enables discrete isolation of T-U wave obscured ectopic atrial activity on the surface ECG while retaining the intricate spatial detail in P wave morphology. Future clinical application of the algorithm may enable improved ECG localization of focal triggers of paroxysmal atrial fibrillation, atrial tachycardia, and the atrial insertion of accessory pathways.