Controllability of the EEG Human Connectome as derived from Phase-Locking Value and Cross-correlation Coefficient

The human brain can be fruitfully represented as a dynamic network according to a balanced scheme of local and global functional and structural connections among brain regions, the so-called connectome, supporting the execution of cognitive and motor functions. Resting-state electroencephalography (EEG) functional connectivity could be exploited to assess the functional organization of the human connectome providing a meaningful description of the functional connections among brain nodes. Moreover, network control theory (NCT) offers a valid framework for modelling the human connectome as a dynamic system and for featuring relevant aspects of the neural activity in terms of regional controllability. However, the signal processing and the consequent choice of a statistical coupling method could affect the estimation of the EEG functional connectome. This study aims to evaluate the agreement between NCT-related controllability metrics as obtained from two different statistical coupling approaches defining the functional coupling between broad-band regional EEG sources, i.e., cross-correlation coefficient (CC) and phase-locking value (PLV). Ten healthy subjects were enrolled and a 10 minutes closed-eyes resting state acquisition was performed. Coefficient of Variation was used to assess the intra-subject variability and Wilcoxon signed-rank test, Passing-Bablok linear regression and Bland-Altman analysis were performed to assess the agreement of the average controllability (AC) and modal controllability (MC) as evaluated via CC and PLV. Results showed lower variability in controllability metrics estimated via PLV. An excellent agreement between the two statistical coupling methods was found for the MC metric, but not for the AC metric. Further analysis on larger datasets and across different canonical frequency bands of EEG signals may possibly reduce the intra-subject variability and/or increase the agreement among NCT-related metrics thereby improving the consistency of the statistical coupling approaches in the estimation of the EEG-derived human connectome.