Chapter 6 – EEG imaging 201 beta (13–30 Hz) and gamma (30–45 Hz). To quantify local neural dynamics associated with resting-state brain activity and pain responses in our nocebo paradigm, spectral power was computed for all EEG electrodes using the Welch method implemented in Matlab. Relative power was computed as the relative contribution of power within a narrow band to the integrated power within the range 1–45 Hz. To investigate whether temporal structure of the EEG changed at rest and during pain responses in our nocebo paradigm, the amplitude envelope was extracted using the Hilbert transform and DFA was computed to quantify LRTC of neuronal oscillations 20–22. DFA quantifies the rate at which auto-correlations of amplitude modulations decay within a signal, with the power-law exponent ranging from 0.5 (uncorrelated) to 1.0 (strong auto-correlations). Signals were filtered using a FIR-filter with a Hamming window with a length corresponding to two f1 Hz cycles for a given frequency band [f1, f2]. To minimize artificial auto-correlations introduced by the FIR-filter, DFA was fitted in the interval from 4 to 20 seconds for alpha band and 2 to 20 secs for beta and gamma bands 20. EEG processing MATLAB 2020a (The MathWorks Inc., Natick, MA, 2014) was used for EEG preprocessing and analysis. Continuous EEG recordings were imported and preprocessed using EEGLAB 35, and analyzed using custom-made scripts from a MATLAB toolbox developed at Vrije Universiteit Amsterdam (VU). All signals were visually inspected for artifacts in windows of 10 seconds. Noisy channels (e.g., with no or bad conductance to the scalp) and segments containing transient artifacts were removed. Next, recordings were re-referenced to the average BioSemi reference. Independent Component Analysis (ICA) was used to project signals to components that are maximally independent from each other 36,37. Eye components were rejected. Continuous EEG
RkJQdWJsaXNoZXIy MTk4NDMw