Application of independent component analysis in atrial fibrillation detection
Na Tang · Journal of Zhejiang University(Engineering Science) · 2006
A non-invasive diagnosis approach for atrial fibrillation(AF) was proposed by extracting atrial activity(AA) signal from real ambulatory electrocardiogram records and analyzing the AA features.Independent component analysis(ICA) theory was used to verify that three fundamental requirements which must be satisfied in ICA are source independence,at most a Gaussian source and instant linear mixture.A mathematical model was formulated for this blind source separation problem.Then a fast and efficient fixed-point ICA algorithm was applied to analyze the simulation and clinical data,and the kurtosis values of separated independent components were also evaluated.The results qualitatively and quantitatively show that the proposed ICA method for AF signal feature extraction and analysis is appropriate and robust.