Adaptive time-frequency parametrization in pharmaco EEG

Abstract

Adaptive time-frequency approximations offer description of the local structures of a signal in terms of their time and frequency coordinates, widths and amplitudes. These parameters can then be used to select and study electroencephalogram (EEG) structures like sleep spindles or slow wave activity (SWA) with high resolution. Such a detailed description of relevant structures improves on the sensitivity of the traditionally used spectral power estimates and opens new possibilities of investigation. These advantages are illustrated using a double-blind test of the influence of zolpidem and midazolam on sleep EEG, and the results are compared with the traditional approach. The observed decrease of frequency of the SWA under the influence of sleep-inducing drugs gives an example of an effect elusive to classical methodology.