EEG acquisition, processing, and spectral analysis

EEG were digitally recorded through a commercially available clinical EEG system with a sampling rate of 256 Hz (Galileo.NET, BE Light system, EB Neuro S.p.A., Firenze, Italy). Each recording lasted at least 20 min in line with International Federation of Clinical Neurophysiology recommendations for clinical EEGs [43]. Electrodes were positioned according to the 10–20 system. Preprocessing and analysis of EEG data were performed using FieldTrip, an open source software package that is implemented as MATLAB® toolbox and enables a broad spectrum of simple and advanced EEG analyses including source reconstruction and network analyses [44]. Standardized epochs of provocation (hyperventilation, photo stimulation) and eyes-open/eyes-closed maneuvers were excluded. Auricular electrodes were removed from the analysis. All channels were referenced to common average. Data preprocessing included detrending and application of discrete Fourier transform filtering at 50 Hz and its harmonics. Data was segmented into artifact-free trials of 10,000 ms. Trials containing artifacts were rejected semiautomatically. Trials containing excessive variance defined by a z-value threshold of 20 were removed. Remaining trials were then visually inspected for artifacts by an experienced EEG reader and rejected as necessary. EEG data was subsequently transformed to frequency space including frequencies of interest between 1 and 70 Hz using a multitaper method fast Fourier transform based on discrete prolate spheroidal sequences windowing. Grand averages were calculated for all trials of one subject. Subject-specific power spectra were standardized by their mean power to allow for comparison of frequency- and sensor-specific power distributions between subjects and groups.