Robust detection of QRS complex using klauder wavelets

Abstract

The level of performance as well as the timing accuracy in the detection of QRS complex may be crucial for further nonlinear biomedical signal processing applied on the derived RR time series. By studying the fatigue process during exercise, a highly non-stationary noise is expected as well as time-varying QRS complex morphology. In difficult recording conditions, the QRS complex detectors have to be extremely robust but also have to meet high performance requirements. To achieve this aim, a “smoothed” matched filtering is realized on specific wavelet coefficient patterns in the time-scale plane. The detection is enhanced using the family of Klauder wavelets which demonstrate similarities to ECG waveforms. The algorithm only needs to capture a correct QRS complex snapshot before launching the detection.