Spatiotemporal Separation Algorithm for Atrial Fibrillation Signal Estimation
Francisco Castells, José Joaquín Rieta, José Millet, David Moratal · 2013
Abstract A new algorithm for the estimation of the atrial activity (AA) signal from the surface electrocardiogram during atrial fibrillation (AF) episodes is proposed. This algorithm has been based on blind source separation methods, and has been designed including statistical and temporal source information. With this new methodology, the estimation of the AA source is enhanced, comparing to the estimation obtained with other independent component analysis (ICA) algorithms. The validity of the proposed method is demonstrated with the analysis of 10 AF recordings. A performance parameter based on the spectral concentration of the AA around its main frequency was employed. The spatiotemporal analysis provided AA signals with an spectral concentration that was in average 13 % above those obtained with ICA methods.