Time-frequency and time-scale methods in the detection and classification of non-stationarities in human physiological data

In this paper, a variety of time-frequency and time-scale methods has been applied to the detection of transients in the heart rate and respiration of neonates during quiet sleep. The methods used include the well-known short-term Fourier Transform, the Wigner-Ville distribution and the wavelet transform as well as the lesser-known instantaneous power spectrum and the Choi-Williams distribution. As far as the authors were aware, the latter two methods have not yet been applied to physiological signals. This study was carried out as part of a project to study the dynamic interactions in the cardiorespiratory system during transient phenomena