Computational protocol: Repetitive transcranial magnetic stimulation induces oscillatory power changes in chronic tinnitus

Similar protocols

Protocol publication

[…] After recording, EEG data were filtered with a high-pass filter of 1 Hz and a low-pass filter of 45 Hz and segmented into epochs of 2 s skipping the first and last two segments of the recording. Thereafter segments were visually inspected for distinct and visible aberrations from the overall recording to identify muscle artifacts (short high frequency oscillations), single channels with low signal-to-noise ratio (zero line in the EEG, main hums, or not smooth trajectories), or other large amplitude physiological artifacts (movement artifacts). Single segments were refused. The data excluding electrodes with low signal-to-noise ratio (maximum of five per measurement) was then subjected to an infomax independent component analysis in order to identify artifact components. Horizontal and vertical eye movement artifact components were removed and the remaining components were back-projected to the EEG signal space. Finally, the data was carefully visually inspected a second time for any remaining artifacts. Thereafter the data was re-referenced to an average reference, the online-reference FCz was reconstructed, and electrodes with signal loss were interpolated.After preprocessing which was done with EEGLAB (Delorme and Makeig, ), data were converted into Fieldtrip format (Oostenveld et al., ) for power spectrum analysis. For power analyses we used the minimal number of available epochs of each measurement (all subjects and all conditions) and therefore chose the first 79 epochs of each measurement. After the fast fourier transformation using a hanning window (Fieldtrip parameters mtmfft and hanning) the power spectrum of each channel, condition, and subject was normalized by dividing the power of each frequency bin by the mean power of the whole power spectrum. The first EEG recording in session 1 and 2 served as baseline condition. For the baseline and sham conditions the two available measurements (one from session 1, one from session 2) were averaged.We were interested in the rTMS induced changes between patients and controls. Thus, data of baseline measurements were subtracted from the post-stimulation conditions as indicators for rTMS induced changes (sham minus baseline, left-frontal minus baseline, etc.) (a comparable statistical approach can be found in Lorenz et al., ). In a second step, we substracted the baseline-corrected data of the sham condition from the baseline-corrected data for the different active conditions (stimulation site specific changes minus sham-induced change). These baseline- and sham-controlled data of patients and controls were compared in an unpaired t-test using a non-parametric permutation test (Fieldtrip parameter montecarlo) with 1000 iterations and a cluster correction to control for alpha inflation due to multiple testing of 63 electrodes. In other words we calculated the following t-test for each stimulation site: patients [(verum-baseline)-(sham-baseline)] vs. controls [(verum-baseline)-(sham-baseline)]. These contrasts were done to identify frequency-specific effects of the four stimulated cortical sites by repeating the t-tests for a priori defined frequency bands (delta: 2–3.5 Hz; theta: 4–7.5 Hz; alpha1: 8–10 Hz; alpha2: 10.5–12.5 Hz; beta1: 13–18 Hz; beta2: 18.5–21 Hz; beta3: 21.5–30 Hz; gamma: 30.5–44 Hz) as suggested by former studies (Vanneste et al., ).Averaged baseline-corrected data of significant clusters were exported into SPSS 22 (IBM Inc., USA) and were analyzed by 2 × 2 analyses of variance (ANOVAs) with the within-subjects factor stimulation protocol (verum rTMS intervention vs. sham) and the between-subjects factor group (patients vs. controls). This was repeated four times according to the four active stimulation sites. To control for group differences in hyperacusis, depressivity, years of education and intelligence these variables were used as covariates. Only effects which were significant for the ANOVA with and without covariates are reported here. In case of significant interaction effects in the ANOVA, post hoc Student t-tests were done. For the illustration of the results exact statistical values were obtained from the SPSS analyses and heat brain maps were generated by Fieldtrip using t-values of the group contrast for baseline- and sham-controlled data. If not otherwise specified default values for data pre-processing and analyses were used. As we were interested especially in rTMS induced changes in EEG power we did not present data with respect to group differences for baseline resting state EEG.For reasons of completeness, baseline and baseline-corrected EEG power for the frequency bands and for both groups for the different stimulation conditions are presented in Supplementary Figures , . […]

Pipeline specifications

Software tools EEGLAB, FieldTrip
Application Clinical electrophysiology
Organisms Homo sapiens