EEG-Based Functional Connectivity Representation using Phase Locking Value for Brain Network Based Applications.

In more recent times, the network perspective study of the human brain has expanded enormously due to the advancements in the field of network neuroscience. Existing methods to form the functional connectivity from the multichannel EEG leads to a fully connected network. Representation of a fully connected functional network with a significant functional network (SFN) can help to characterize and quantify the complex brain networks. Further, it can also provide novel insight into the brain cognition analysis and is crucial in several brain network-based applications. This paper presents a framework to find the SFN corresponding to any event from its fully connected network. Using the phase-locking value (PLV) in EEG we first identify the difference PLV of an event to the rest. Based on the difference PLV, we then identify the reactive band and event-associated most reactive pairs (MRPs). The SFNs corresponding to their events is then formed with the identified MRPs. The proposed method is employed on 'database for emotion analysis using physiological signals (DEAP)' data set to find the SFNs associated with emotions. Comparable state-of-the-art multiple emotion classification accuracies are obtained with the identified SFNs. Results show that the proposed methods can be used as a general thresholding technique to identify the event-related SFNs which are crucial in brain network-based applications.