Research On The Network Mechanism Of P300

P300 is an endogenous event-related potential that is closely related to brain cognitive function.P300 is not only widely applied in cognitive neuroscience and brain-computer interface(BCI),but also becomes an indispensable research tool of the clinical diseases,e.g.,the clinical diagnosis of schizophrenia(SCZ).However,the current researches on the neural mechanism of P300 and the differences between subjects still have limitations.In our present study,based on the electroencephalogram(EEG)and functional magnetic resonance imaging(fMRI)techniques,using the brain network and graph theory analysis as the primary research tools,we then systematically explore the neural mechanism of P300 from different modalities and levels of brain networks by investigating the brain activity of both resting and task states.We hope our current work can achieve a better neural theoretical basis subserving its application to multiple research areas.The main contribution of this dissertation consists of the following five aspects:1)From the perspective of the resting-state EEG network,we first investigate the P300 evoked in the Oddball task and the potential network mechanism accounting for its information processing.The results show the P300 amplitudes are positively correlated with the mean functional connectivity,clustering coefficient,global efficiency,and local efficiency of the resting-state EEG brain network,but negatively correlated with the characteristic path length,in the meantime,the network structure positively correlating with the P300 amplitude is distributed at the prefrontal/frontal and parietal/occipital lobe.The above results indicate that efficient resource allocations in the prefrontal/frontal and parietal/occipital lobes at rest are closely correlated with the activation of the task P300.2)Within different rhythms,how does the brain switch from a resting state to a task state,how related brain resources are integrated,and how corresponding brain networks are reorganized are still left unclear.We investigate the network mechanism of P300 by further uncovering the reconfigured EEG network architectures during the transition from resting state to P300 task.The results indicate the P300 amplitude presents a significantly positive relationship with the reconfigured network coupling strength in 1-8 Hz,but a significantly negative relationship with that in 8-13 Hz.Moreover,the classifications based on the reconfigured network index of both bands imply this index can distinguish the resting and task states with 100% accuracy and the high-and low-amplitude groups with an accuracy of 77.78%.The work elucidates that the reconfiguration pattern of the brain network could reflect the brain efficiency related to the P300 information processing,which usually switches within different rhythms in a dynamically balanced way.3)Based on the simultaneous EEG-fMRI techniques,by focusing on a more subtle(millimeter)brain structure,we investigate the reconfigured pattern of task-activated fMRI network from resting to task state and its relationship with the concurrent EEG-P300,as well.The results show that the identified 14 task-activated regions of interest(ROIs)and the fMRI network based on these ROIs are closely correlated with the P300 amplitudes;and compared to the subjects with higher P300 amplitudes,when the brain transits from resting to task state,the subjects with lower amplitudes experienced more enhanced reconfigured interactions among related ROIs.The above results show obvious inter-individual variability of the brain resource reallocation and network reorganization related to the P300,and the higher the network reconfiguration capability,the larger the task information processing efficiency.4)The stable brain network analysis fails to capture the synchronized interactions among multiple brain regions on the millisecond scale.Herein,based on the adaptive directed transfer function(ADTF),we further investigate the dynamic information processing of the P300 by mining its time-varying networks.The results show that concerning the decision process stage,the central area of the brain serves as the control source,while during the neural response stage,the right prefrontal lobe becomes the new control source,and the top-down information flow from the right prefrontal lobe to the posterior parietal lobe controls the neural responses to the task stimuli.This part illustrates that the task elicitation of the P300 involves different stages of dynamic information processing,which corresponds to distinct time-varying brain network structures.5)Although the P300 is demonstrated to be a potential diagnostic biomarker for SCZ given its close association with attention and working memory,its waveform features still cannot achieve satisfying diagnosis performance.To further improve the clinical diagnosis of the SCZ,based on the resting and task P300 datasets,we adopt the P300 amplitudes,network properties,and spatial pattern of the network(SPN)features to accomplish the clinical diagnosis of the SCZ.The results show that compared with the other two types of features(i.e.,P300 amplitude and network properties),the SPN features extracted from related brain networks achieved better diagnosis performance when distinguishing the SCZ patients from the healthy controls.Moreover,the fused SPN features of different brain states could achieve the highest classification accuracy of 90.48%.This part finally elucidates the differences of the P300 and its networks between the SCZ and healthy controls,which provides comprehensive information for the understanding of the SCZ pathogenesis and also indicates that the SPN has great potential for the diagnosis of clinical diseases.To sum up,on one hand,based on resting and task brain network and its potential reconfiguration pattern when transiting from resting to task state,this paper explores the neural mechanism that accounts for the P300 information processing;on the other hand,based on the features of resting and task brain network topologies,the accurate diagnosis of schizophrenia is realized by combining the multi-state and multi-feature fusion strategies.The above work may help promote the practical application of P300 in clinical disease diagnosis and the other fields,as well.