Convolutive Multiband Blind Separation to dissociate atrial from ventricular activity in atrial fibrillation

In order to use the ECG as a tool for the characterization of atrial fibrillation (AF), we need to dissociate atrial activity (AA) from ventricular activity. On the other hand, the reduced number of leads recorded from a Holter system limits the necessary spatial diversity required by Blind Source Separation (BSS) techniques to accurately extract the AA. In this work, we propose a new method, the Convolutive Multiband Blind Separation (CMBS), to solve the problem of reduced number of leads by combining the Wavelet transform with the convolutive BSS algorithm In-fomax. Our analysis shows up that CMBS improves the extraction performance of AA from Holter systems in comparison with previous extraction methods. This improvement is accomplished in two different scenarios, one for synthetic signals and another one for real signals. A high accuracy of the estimated AA for synthetic and real AF ECG episodes is reached in both scenarios. In addition, results prove that CMBS preserves the original AA spectral parameters.