A Research On The Method And Application Of EEG Chronological Information Mining During Naturalistic Stimuli

Naturalistic stimuli is a kind of stimulus that lasts a long time and fits the actual life scene.Compared with traditional experimental stimulation,it is closer to people’s daily life,and can provide richer stimulation to study brain function activities.Among them,Action-video Game(AVG),as a typical naturalistic stimuli stimulus,is widely used in cognitive neuroscience research.Electroencephalogram(EEG),characterized by its portability and ease of collection,is especially suitable for the study of brain function stimulated by natural patterns such as action video games.However,most of the current traditional EEG studies focus on the spatial characteristic information(such as EEG power spectrum,amplitude,etc.)of the task state or the resting state or the EEG functional connection in the time history,and lack to fully explore the information of continuous EEG in the time-spatial domain at the same time,so that when faced with more flexible and rich naturalistic stimuli,it is difficult to effectively reflect the complex cognitive activities of the brain on the spatial and temporal scale.Therefore,aiming at the problem of information mining of the process of brain functional activities stimulated by naturalistic stimuli,this thesis makes full use of the characteristics of high temporal resolution of EEG,and proposes a new method of EEG chronological information mining to realize the in-depth analysis of the activity information of brain functional activities in spatio-temporal scale.Further,it is applied to the research of brain mechanism related to action video games,which provides a new perspective for us to understand the mechanism of complex brain function.Specific work is as follows:(1)Aiming at the high temporal resolution of EEG and the characteristics of naturalistic stimuli stimulation(long duration,dynamic and unstable),an effective method of mining EEG chronological information was proposed by using the highintensity activity of specific EEG frequency band to define chronological characteristic events.Furthermore,the rest state data set(data set 1,68 cases)was used to determine the reasonable event threshold,power spectrum analysis window length and other important parameters.Finally,the validity of the method was preliminarily verified by using the rest state data of open and closed eyes(data set 3,39 pairs of 78 cases in total).The results showed that,under certain parameters,the chronological information network was in a relatively stable state,and the eye-opening and eye-closing resting states had significant differences in Delta,Theta and Alpha bands(FDR,p<0.05).There were significant differences in the conventional brain networks in the Theta,Alpha and Gamma bands(FDR,p<0.05).The results of the chronological information network are consistent with previous studies and reflect the new differences in temporal and spatial information of the open and closed resting EEG.The above results show that the chronological feature information mining method proposed in this thesis can effectively mine EEG temporal and spatial information at the same time,and has good robustness.(2)Further,the above method and timing information index were applied to the ecological game data set(Data 2,League of Legends games,65 cases of data;Data set4,66 cases)to investigate the correlation between this index and behavioral data,as well as the differences of this index in game state among subjects with different game levels.The results showed that brain timing information was significantly correlated with visual search(response time)in Delta,Theta,Alpha and Beta bands(p<0.05),and was significantly correlated with visual tracking(response time)in Alpha bands(p<0.05).The timing information connection of skilled AVG players in Alpha band was significantly higher than that of novice AVG players(FDR,p<0.05).Finally,based on the analysis results of EEG timing information of behavioral correlation and different game levels,the brain function changes under game state were revealed,and the brain function mechanism related to action video games was explored.In conclusion,the work in this thesis has developed and preliminarily perfected a chronological information mining method for EEG data,which can provide a new reference for information mining of EEG data stimulated by naturalistic stimuli including AVG in the future.