Two stage SVD-based method for QRST waves cancellation in atrial fibrillation detection

The paper presents the results of studies devoted to analysis of performance of two-stage SVD-based (singular value decomposition) method for cancellation of QRST waves in ECG (electrocardiogram) signals. In our studies, the purpose of cancellation of QRST (related to ventricular activity) is to extract the main atrial activity (AA) from ECG signal. A special feature of the analyzed method is SVD-based estimation and subtraction of QRS-complexes and T-waves in two consequent stages instead of one. To improve the quality of QRST cancellation we proposed upsampling of ECG signal to increase the time resolution. The results of preliminary experiments show that two stage SVD-based method with appropriately chosen parameters can allow achieving better performance than the method based on single-stage SVD-based QRST cancellation, in applications related to atrial activity analysis, e.g. detection of atrial fibrillation (AFib) symptoms on the basis of atrial activity.