Research on f wave extraction methods of body surface potential mapping recordings from patients with atrial fibrillation

Wang Dexi,Yang Cuiwei

DOI: https://doi.org/10.19650/j.cnki.cjsi.2016.10.024

2016-01-01

Abstract:Body surface potential mapping (BSPM)is a noninvasive electrocardiographic technique and gains more and more applications in the study of atrial fibrillation.BSPMrecordings of the patients with atrial fibrillation can be modelled as an instant linear mixture of three statistically independent source signals of ventricular and atrial activities as well as noise signal.In order to extract the atrial activity (f wave)from BSPMrecordings,two kinds of blind signal separation methods were applied,which are (ICA)method and the combination of ICA and (SOBI)called ICA-SOBI method.The two methods were applied on nine patients with atrial fibrillation to extract the f wave from the BSPM recordings.The spectral concentration (SC)was adopted as the evaluation index to evaluate the extracted result.Statistical result shows that the f waves extracted with ICA-SOBI method have higher SC and the extraction effect is much better.Meanwhile,the ICA-SOBI method was used to extract the f waves from the BSPMrecordings at different sub-regions.The f wave extracted from the region nearest to the heart shows the highest SC;and this result can be a guidance for the spatial distribution design of the electrodes for BSPM.