Research On Language Laterality Based On Clustering Methods In Source Space With Resting-state EEG/MEG

In recent years, research on resting-state brain functional network has become an importantpart of neuroimaging. The resting state can reflect the intrinsic brain’s functional architecture. Thefunctional model of resting-state has related to the evolution of the brain plasticity, which will alsogenerate the lateralization affectting the brain pattern. As one of the most important advancedfunctions of brain, language is a tool of thinking and communication. The language function iscontrolled not only by a particular cerebral cortex, but also the cooperation of several brain regions,furthermore, the contributed regions is different.In this thesis, we focus on extracting the lateralization with the resting-state EEG/MEG datawhich can be combined with task state or even substitute the task state to reveal the phenomenonof lateralization. Firstly, because EEG and MEG are high temporal resolution, but less spatialresolution, the Beamformer algorithm was applied to reconstruct the electrical activities in soucespace for the improvement of spatial resolution. Secondly, methods of the envelope energycorrelation and directed transfer function were adopted to inspect the functional and effectiveconnectivity for different frequency band, respectively. Finally, K-means clustering algorithm wasadopted for pacellating language area based on the results of functional connectivity to illustratethe language lateraliztion.It shows that the brain displays the left laterality under different frequency bands for languageareas, and the language laterality is mainly caused by the right brain activation instead of the leftbrain through comparison of different subjects and different brain regions. In summary, thelateralization phenomenon of brain can be detected from the resting state data, which can befurther applied to research the change of brain functional network and lateralization for epilepticand brain damaged subjects.