A wavelet methodology for EEG time-frequency analysis in a time discrimination task

EEG signals recorded by surface electrodes placed on the scalp can be thought as nonstationary stochastic processes in both time and space, especially in response to external stimuli. Cognitive tasks, in particular, are reflected by changes in EEG dynamics concerning both rhythms energy and connectivity across different brain regions. In the frequency-domain, EEG analysis is complicated and time-frequency methodologies are needed. The Wavelet Transform, in particular, represents a powerful tool for analysing, within a time-frequency embedding, the EEG. In this study we applied a wavelet-based methodology to extract quantitative time-frequency parameters from EEG signals recorded during a time discrimination task in 12 subjects. We used a continous wavelet transform with a complex Morlet as mother function. In order to improve the time-frequency resolution and to make it satisfactory, each of the four standard EEG rhythms (i.e. theta, alpha, beta, gamma) was studied with Morlet wavelet parameters tuned ad hoc on the basis of both the width of the specific frequency band and the particular type of activity under examination. The numerical values of the estimated time-frequency indexes were then compared, evidencing statistically significant differences in the brain response between experimental conditions