Time- and Frequency-Domain Analyses of Auditory Evoked Fields

Abstract

The magnetoencephalogram (MEG) represents a superposition of magnetic field contributions from a great number of individual generators with different temporal activation patterns and different frequency characteristics. Thus it is not surprising that the result of a source analysis is generally dependent on the selected time and freqency windows. While the time dependency of the magnetic field distribution is routinely analysed in most MEG studies, investigations of the frequency dependency are still the exception. Two different concepts for a frequency-specific source analysis can be distinguished. First, after narrow-band filtering of the data a time-domain source analysis can be performed in the usual way. Second, special frequency-domain source analysis procedures can be applied to the Fourier transformed data. The first procedure was used e.g. by Williamson et al. (1989), Lü et al. (1992), and Grummich et al. (1992) for the investigation of human parietooccipital alpha activity and by Pantev et al. (1991, 1993) for the investigation of gamma-band auditory evoked fields. The second procedure was proposed already years ago (Lehmann and Michel, 1989, 1990; Michel et al., 1992, 1993) for the analysis of the electroencephalogram (EEG). Inspired by these studies, Lütkenhöner (1992a) formulated a generalized theory and emphasized the close relationship between a source analysis in the time- and the frequency-domain. Recently, Tesche and Kajola (1993) used a frequency-domain analysis to study magnetic alpha activity as well as slow wave activity with superimposed spikes.