Multidimensional ICA for the separation of atrial and ventricular activities from single lead ECGs in paroxysmal atrial fibrillation episodes

Abstract

The analysis of paroxysmal atrial fibrillation requires the previous estimation of the atrial activity (AA) from one lead ECG. Considering the statistical properties of the cardiac electrical activities, it follows that both AA and ventricular activity (VA) present a high redundancy degree at different time intervals, whereas AA keeps independent from VA. This contribution adopts a multidimensional independent component analysis (MICA) formulation in order to find a set of components that minimises the mutual information existing in the ECG signal at different intervals. The independent components can be grouped in VA and AA subspaces, what enables the reconstruction of the AA at each observation point from the AA subspace. The proposed approach is validated with a significant database composed of simulated and real AF recordings. © Springer-Verlag 2004.