Figure 2

Full process pipeline. The EEG data underwent two preprocessing steps, namely EEG Preprocessing Step I (blue box) and EEG Preprocessing Step II (green box). Independent Components (ICs) were extracted within Step I. IC weights were then transferred to the dataset, conservatively processed according to Step II. Step I is designed to extract ICs with the best possible reliability. Step II is more conservative, and it is designed to retain the maximum amount of information for subsequent connectivity analyses. Each IC was source-localized using a dipolar model, and artifact clusters of ICs were removed from the data. The backprojected EEG data was then used to perform distributed source localization, and the AAL atlas EEG ROIs time courses were extracted. Cortico-cortical connectivity analysis was then performed across all subjects for 0 GF, 100 GF, and Rest conditions using the time course of the source-localized activity within the regions of interest (ROIs) extracted after Step II. In parallel, EMG data were aligned to EEG, high-passed at 2 Hz and noisy windows labeled for rejection. Only time windows that were not labeled as artifact either for EEG or EMG were retained. Cortico-muscular connectivity (assessed through Partial Directed Coherence—PDC) was estimated for 0 GF and 100 GF respectively on epoched data, time locked to the gait cycle (see “Methods” for more details). Group-level statistical analysis across all subjects and conditions was then performed.