Investigation of physiological and pathological conditions using electroencephalographic connectivity metrics

Functional connectivity (FC) metrics identify statistical (undirected) associations among distinct brain areas and therefore represent a powerful tool to investigate brain inter-regional interactions in distinct behavioural states. However, the application and interpretation of FC in electrophysiological data is impacted by important confounds related to the instantaneous propagation of electric fields generated by primary current sources to many of the on-scalp sensors – the so-called phenomenon of “volume conduction”. Because of this linear mixing of different sources, common FC methods may lead to the identification of apparent couplings that do not reflect true brain inter-regional interactions. To overcome this problem, new FC metrics have been specifically designed to minimize the impact of volume conduction. Among these novel methods, the weighted Phase Lag Index (wPLI) and the weighted Symbolic Mutual Information (wSMI) attracted a growing interest during the last decade, and have been successfully applied to describe brain function in a wide range of different conditions, including states associated with altered levels of consciousness. In spite of the many promising applications and results, the two methods have never been characterized in detail, nor compared to investigate their potential similarities or differences. Given these premises, in the present thesis, my aim was to assess the properties of wPLI and wSMI in order to define their respective potential advantages and disadvantages, as well as to determine whether useful information could be gained through their combined application. To this aim I performed three distinct, complementary studies. In my first project, I simulated realistic high-density EEG data based on imposed interaction dynamics between sources of interest to test the accuracy of wPLI and wSMI at detecting different types of linear and nonlinear functional interactions. Based on the resulting finding that they provide complementary information, I applied the two methods to the study of EEG data, collected in physiological and pathological states. In my second study, I analyzed power, wPLI and wSMI changes across distinct physiological stages of vigilance, specifically wakefulness (W), NREM and REM sleep in 24 healthy participants. Specifically, I explored the role of power- and FC-based features in identifying differences between all stages of interest (W, N2, N3, REM), stages characterized by higher (W+REM) and lower (N2+N3) probabilities of conscious experiences and differences in sensory disconnection (REM vs. W), using a cross-participant classification paradigm. Finally, in my third study, I applied the two methods for investigating the effects of motor rehabilitation on brain functional correlates in 16 multiple sclerosis patients. Obtained results demonstrated that wPLI and wSMI provide distinct and complementary information about functional brain dynamics and indicate that the conjoint use of these two methods may represent a powerful tool to investigate brain connectivity in physiological and pathological conditions.