Component selection for Principal Component Analysis-based extraction of atrial fibrillation

For the study of atrial fibrillation (AF) in the surface ECG, the cancellation of the QRS-T is required in order to isolate the atrial from the ventricular activity. Principal Component Analysis (PCA) was previously employed with good results. The main problem with this method is the selection of the principal components that contains the AF wave information. This paper presents a study to determine the best subset of the 12 principal components computed from a 12 lead standard surface ECG in order to optimize performance. A test database consisting of 840 ECGs with simulated AF was developed. This test dataset was used to determine the performance of the PCA when retaining different subsets of the principal components. It was observed that the components 3 to 8 contributed mainly to the atrial fibrillation wave. Finally, the best PCA variant found was used to analyse the PTB AF database. The distribution of the main frequencies and the concentration of the spectral energy around the main frequencies were determined for this database..