Feasibility of separating the atrial and ventricular components of the electrocardiogram

Our aim was to assess the feasibility of separating concurrent atrial and ventricular (AV) activities of the body surface signals using an algorithm in patients with heart block. Principal Component Analysis (PCA) was applied to 12-lead ECG sections containing 'P wave on T wave' from two patients with complete heart block. The principal components containing the separated atrial activity were identified and the inverse transform applied to project the atrial activity back into the ECG leads. These extracted P waves were then compared with reference P waves occurring without ventricular activity. The extracted and reference P waves were highly correlated (0.7 in patient 1, 0.8 patient 2). PCA was able to separate the AV activities in heart block patients.