Spectral analysis is important in many fields, such as speech, radar and biomedicine. Many signals encountered in these areas possess time-varying spectral characteristics. The power spectrum indicates what frequencies exist in the signal but it does not show when those frequencies occur. Time-frequency analysisprovides this missing information. A time-frequency representation of the signal shows the intensities of the frequencies in the signal at the times they occur, and thus reveals if and how the frequencies of a signal are changing over time.Time-dependent spectral analysis of beat-to-beat variations of cardiac rhythm, or heart rate variability (HRV), represents a major challenge due to the structure of the signal. A number oftime-frequency representations have been proposed for the estimation of the time-dependent spectra. However, time-frequency analysis of multicomponent physiological signals such as cardiac rhythm is complicated by the presence of numerous, ill-structured frequency elements. We sought to develop a simple method for 1)detecting changes in the structure of the HRV signal, 2)segmenting the signal into pseudo-stationary portions, and 3)exposing characteristic patterns of the changes in thetime-frequency plane. The method, referred to as Orthonormal-Basis Partitioning and Time-Frequency Representation (OPTR), is validated on simulated signals and HRV data. Unlike the traditional time-frequency HRV representations, which are usuallyapplied to short segments of signals recorded in controlled conditions, OPTR can be applied to long and "content-rich" ambulatory signals to obtain the signal representation along withits time-varying spectrum. Thus, the proposed approach extends the scope of applications of the time-frequency analysis to all types of HRV signals and to other physiological data.
展开▼