Signal acquisition and analysis
A telemetric Nautilus EEG system (g.tec, Schiedlberg) was used to acquire ECG and EEG signals according to a modified International 10-20 System. In details, leads were placed in F3, F7, C3, T7, P3, PO3, F4, F8, C4, T8, P4, PO4, Fz, Cz, Pz, Oz positions and referred to Cz. Two additional electrodes were placed on the skin near lateral cantus of left eye and in correspondence of left orbital ridge to detect eye movements. Another electrode, referred to Cz, was placed in proximity of left shoulder for ECG signal. All impedances were kept below 5 kΩ.
The ECG signal was analyzed with Kubios HRV (Tarvainen et al., 2014). First, the RR series were extracted from the ECG using the well-known Pan-Tompkins algorithm (Pan & Tompkins, 1985) and a cubic spline interpolation method was used to correct algorithm-related peak-detection artefacts. The obtained RR series were then resampled to 4Hz to derive the HRV signal (Malik, 1996). Starting from the HRV, we extracted several features in the frequency and time domain. Specifically, we computed the mean RR interval duration (RR, ms), the total RR variability (SD) the square root of the mean squared differences of successive normal-to-normal (NN) intervals (RMSSD), the power expressed as a percentage of total power in the low-frequency (LF, 0.04–0.15 Hz) and high frequency (HF, 0.15–0.40 Hz) ranges, the ratio of LF to HF power (LF/HF ratio). R peaks markers were saved for subsequent HEP analysis.
EEG signals were analyzed using EEGLAB (Delorme & Makeig, 2004) and MATLAB 2020b (The Mathworks, Inc., 2020) custom scripts. All EEG signals were downsampled to a sampling frequency of 125Hz after applying a proper low-pass anti-aliasing filter. Then, a high-pass filter of 0.1Hz was applied. Bad channels were removed through a semi-automatic procedure. First, the channels whose correlation coefficient with their neighbors was lower than a predefined threshold were discarded (here set to 0.8) (Mullen et al., 2013). Then, an expert visually inspected the data to eventually identify and remove those bad channels not captured by the correlation criterion. Afterwards, removed channels were recovered using a spline-spherical interpolation method. The obtained signals were re-referenced to the numeric average of all channels before undergoing to independent component analysis (ICA) (Makeig et al., 1995). ICA decomposes the signals into sets of maximally statistically independent components that represent both brain sources and different type of artefacts (e.g., muscular, ocular) (Onton et al., 2006). Artefact-related ICs were identified through visual inspection of their maps, spectra and time-course (Delorme et al., 2012). Each independent component was additionally described through its associated equivalent current dipole (Oostenveld, 2011). Finally, the EEG signals were reconstructed without the contribution of artefact-related ICs (i.e, ocular, muscular, cardiac, channel noise and other sources of artefacts). The HEP amplitude was obtained by averaging the EEG signals synchronized the R-peaks previously obtained with Kubios (Tarvainen et al., 2014). More specifically, we extracted EEG epochs from -200 ms to 600 ms around each R-peak followed by subtractive baseline correction estimated in 200ms preceding the R-peak. The epochs contaminated by abrupt signal changes in the ECG or EEG signal were also discarded from the analysis. At the end of this procedure, we obtained a collection of EEG epochs for each condition for each subject (% of retained epochs > 95%).