Atrial fibrillation wave extraction algorithm based on ICA and wavelet transform

When independent component analysis (ICA) algorithms are used to extract atrial fibrillation (F) wave from electrocardiogram signal (ECGs) of persistent atrial fibrillation, serious distortion exists in F wave QRST segment. To reduce this distortion, a new algorithm based on ICA and wavelet transform is proposed. Firstly, ICA algorithm is applied to get initial F wave and its separation vector. Then wavelet transform is used to decompose initial F wave, and the objective function that reflects F wave distortion is constructed in wavelet domain. Finally, steepest descent method is used to optimize the objective function and get more accurate separation vector. The proposed algorithm is validated with both simulated signal and real ECGs of atrial fibrillation, and distortion in the extracted F wave is considerately reduced.