We applied the EEG source analysis procedure introduced in our previous paper (10). The raw data from EEG sensor signals were projected onto the cortical surface employing individual head models and a linearly constrained minimum variance beamformer method (31) to obtain the source signal in each specific brain region. For each source region, the signal (number of voxels times number of time steps) was ICA decomposed as proposed by Jonmohamadi et al. (32). That is, the resulting components, representing regional source patterns with mutually independent time courses, were ordered according to their explained variances and a minimal set that explained 95% of the data variance was selected. This procedure helped eliminate noise and limit the number of independent sources patterns in each region. This study investigated the four brain regions that we previously found to show statistically enhanced PAC in patients with PD compared with controls (10). These regions were PMC, M1, BA3, and BA1&2, which were defined with reference to the multimodal parcellation by Glasser et al. (33). Details are described in the Supplemental Method “Region-based source analysis.”. Therefore,
