Agreement of Global Topological Properties of EEG Connectomes as Estimated from Phase-Locking Values and Cross-Correlation Coefficients

The human brain can be viewed as a complex system of structurally and functionally interconnected units (nodes) that dynamically evolves through different activation states supporting cognitive functions. Emerging evidence suggests that electroencephalography (EEG) source functional connectivity may offer a valid complementary solution to functional MRI (fMRI) to provide a meaningful description of functional resting state connections among brain nodes. However, signal processing and particularly methods for the statistical coupling between EEG source signals could affect the evaluation of the functional connectivity. The aim of this study was to evaluate the agreement between three global graph theory metrics using two different statistical approaches to define functional coupling between regional EEG sources, i.e., cross-correlation (CC) and phase-locking value (PLV). Ten healthy subjects were enrolled in this study and a 10 minutes eyes-closed resting state acquisition was performed. The acquired EEG signals were processed to compute the functional connectivity matrices and three global parameters of graph theory were calculated: global efficiency, transitivity, and mean participation, describing the level of integration, segregation and modularity of the network, respectively. The agreement of each parameter obtained by using both CC and PLV was assessed by means of the following statistical approaches: non-parametric paired test, Passing-Bablok linear regression and Bland-Altman analysis. Study results showed that there was not perfect agreement between the two methods of statistical coupling for global efficiency and transitivity while for the mean participation a perfect agreement was found. Further analysis on larger datasets could derive more robust results about the agreement among graph theory parameters in order to assess the consistency of different statistical coupling approaches in the evaluation of the functional connectivity matrix from EEG data.