Eeg Of The Brain Network Method And Applied Research

Complex network has been widely applied in the brain function studies. From the synaptic connections linking neuronal units, the interdependent nonlinear dynamic activities between neuron sets, to the causal interaction among brain regions, the brain network represents different levels of brain activity states in terms of structural connectivity, functional connectivity and effective connectivity. The studies of default mode network in resting state, and the event-related brain network induced by task stimulus provide much more information for brain cognitive processing. The electroencephalogram (EEG) technique is ideally suited to study the neural activity on large time scale. With the advance at high temporal resolution and large number of recording sensors covering the whole head, EEG technique is increasingly used to study brain networks. The connectivity can be both mapped on the scalp and the cortex at the source domain through estimating the cortical sources.During the brain network studies, the scalp network must be affected by the reference. It is fundamental to choose a reasonable reference to reconstruct proper brain network. To analyze the errors introduced by different references in brain network configuration, and find the feasible reference type is one of the tasks in this paper. Moreover, as to the event-related network, the accuracy of event-related potential (ERP) components has great effect on the network. In the short inter-stimulus intervals experiment, ERP components may overlap, and the conventional event-evoked averaged EEG may introduce distortion. Therefore, it is crucial to obtain the pure decomposed components before brain network studies. Accordingly, we take the two fundamental issues into the studies of EEG default mode network (DNN) and spatial orientation brain network, and a series of detailed researches has been carried out. Our major works are listed as follows:1. In the simulation, comparing the errors in EEG coherence analysis and the effects on brain network introduced by different references, we concluded the infinity reference based on the reference electrode standardization technique (REST) can generate the smallest error and exactly recover the object brain network. 2. Based on the spectral analysis of the resting state EEG, significant difference was found on the spectral power, spatial distribution, weighed amplitude centre and coherence. The DMN was constructed according to the coherence coefficient in a wide frequency bands with different references, and it is observed that comparing to REST, the other references brought great distortion on DMN.3. A robust decomposition algorithm was used to obtain the decomposed components of the short-stimulus intervals ERPs in spatial orientation experiment. The properties of the spatial orientation ERPs were analyzed and the results demonstrated the decomposition algorithm can efficiently solve the components-overlap problem. The event-related network which was consistent with the neural physiological mechanism was shown according to the synchronization analysis between nodes.