Atrial activity estimation in atrial fibrillation by combining ESC and ICA methods

Atrial fibrillation (AF) is the most prevalent arrhythmia with which the physicians are confronted nowadays. This type of affection can lead from problems like hypertension, to life threatening problems such as thromboembolic events. It can be identified using a routine electrocardiogram (ECG) investigation and is characterized by the chaotic-like oscillations, f-waves, which replace the P wave. In the present paper a combination of two conceptually different methods is presented: the Event Synchronous Canceller, ESC, for the cancellation of the QRS-T and Independent Component Analysis, ICA, for the extraction of the AF signal. The proposed algorithm is applied successfully on both realistic simulated ECG with AF signals and on a database of real signals, “St.-Petersburg Institute of Cardiological Technics 12-lead Arrhythmia Database” from the MIT database Physionet.