Analysis of variability: a system for comparing classical, parametric, adaptive and Wigner-Ville power spectral estimators

The typical approach to the study of variability of physiopathological conditions consists of the analysis of a power spectral density (PSD) function estimation, computed on a time series extracted from the signal of interest. In order to facilitate the correct choice among the available methods for PSD estimation in practical situations, different PSD estimators have been applied extensively to identify their performance. It is shown that with the classical, fast Fourier transform (FFT) based approach, 39 tapering windows can be tested, allowing the computation of the FFT-based spectrograms, with a variety of options. The parametric approach allows different estimators to be tested in different situations. The Wigner-Ville PSD estimator is introduced, which is able to associate a PSD to each time domain sample, allowing an appreciation of the variability of the PSD even in nonstationary data segments. A comprehensive graphical interface allows a visual comparison among the different estimators.<
>