A new constrained Independent Component Analysis method to analyze atrial arrhythmias

An arrhythmia occurs when the rhythm of the heartbeat is altered. Atrial arrhythmias are very common; they appear when electrical signals travel through the atria in a fast and disorganized way. The analysis of atrial arrhythmias include the estimation of the atrial component of the recorded electrocardiogram (ECG). This task can be done by Independent Component Analysis (ICA) since the atrial component is independent from the rest of signals superposed in the recording. We present a new algorithm based on ICA that adds to the independence ICA criterion a restriction about the waveform of the atrial signal. The constrained ICA algorithm allows to solve the inherent permutation ambiguity of ICA algorithms, since the first recover signal is the one close to the template incorporated to the separation process. The results confirm the goodness of this new algorithm since the ventricular component of the signal is canceled.