A two-stage algorithm for extracting atrial fibrillation signal

In this paper, we propose a two-stage blind source extraction (BSE) based algorithm for extracting an atrial fibrillation (AF) signal, which exploits both spatial and temporal information contained in the recorded ECG signals. In the first stage, the statistical independence, the high-order statistics, and AF's kurtosis value range are exploited to roughly extract an AF signal. In the second stage, the extracted AF signal is further processed as cleanly as possible, based on its temporally correlated and quasi-periodic property. The algorithm can directly extract only one clean AF signal while blind source separation (BSS) methods and primary component analysis (PCA) algorithm obtain multi signals and then judge which one of the separated multisignals is an AF signal. Therefore, our proposed algorithm can save lots of time and resources. Extensive real ECGs experiments have confirmed that it is an appealing method in clinical diagnosis.