AUTHOR=Huang Yunzhi , Zhang Junpeng , Cui Yuan , Yang Gang , Liu Qi , Yin Guangfu 

TITLE=Sensor Level Functional Connectivity Topography Comparison Between Different References Based EEG and MEG

JOURNAL=Frontiers in Behavioral Neuroscience

VOLUME=Volume 12 - 2018

YEAR=2018

URL=https://www.frontiersin.org/journals/behavioral-neuroscience/articles/10.3389/fnbeh.2018.00096

DOI=10.3389/fnbeh.2018.00096

ISSN=1662-5153

ABSTRACT=Sensor level functional connectivity topography (sFCT) contributes significantly in reflecting underlying brain networks. Two different modalities, magnetoencephalography (MEG) and electroencephalography (EEG), can be utilized to construct the sFCT. In this study, intra-modality and inter-modality comparisons of sFCT are investigated. The intra-modality refers to the comparison of sFCT between different EEG references, including the Reference Electrode Standardization Technique (REST), average reference (AR), linked mastoids (LM) and left mastoid references (LR). The inter-modality means that the comparison of sFCT is performed between MEG and EEG, the two different bioelectromagnetic modalities. Three metrics are exploited to quantize the difference of FCTs, including Relative Error (RE), Overlap Rate (OR) and Hamming Distance (HD). The more similar on sFCT, the less RE and HD are, while the higher OR is. The simulations showed that, for intra modality comparisons, REST can decrease the reference effects on scalp EEG recordings and thus, REST based sFCT could obtain results closer to the ground truth (sFCT based on ideal recordings). While for inter modality comparisons, among all the reference schemes, REST based sFCT is most similar with that derived from MEG, which is simultaneously recorded with EEG. In addition, simultaneously recorded MEG and EEG from publicly available face-recognition experiments were analysis using similar pipeline as in the simulations. The result showed 1) if taking MEG sFCT as standard, REST based sFCT, along with LM based sFCT, obtained results closer to “standard” in the terms of HD; 2) REST based sFCT and MEG sFCT had highest similarity in the terms of RE; 3) REST based sFCT had most overlapping edges with MEG sFCT (in the terms of OR). This study provided new insights into 1) the effect of different reference schemes on sFCT; 2) the commonality between MEG and EEG in the terms of sFCT.