Time-Frequency Filtering with the S-Transform of ECG Signals

A key feature of the S-transform is that it uniquely combines a frequency dependent resolution of the time-frequency space and absolutely referenced local phase information. The S-transform is a time-frequency representation known for its local spectral phase properties. In this paper, a method to process non-stationary signal, such as electrocardiograms (ECG) based on S- transform, in which a filter is applied to a time frequency distribution instead of the Fourier spectrum. Such distribution is the S- transform, a modified short-time Fourier transform whose window scales with frequency, as in wavelets.