Research On Control Strategies And Microstate Dynamics Of EEG Brain Networks Based On Computational Models

According to the statistical analysis from the World Health Organization,brain diseases have become the largest burden that seriously affects the well-being of human beings.It is a very important field in brain science to determine their pathogenic mechanism and early diagnosis indicators,and to develop new technologies for corresponding intervention,treatment and rehabilitation.Schizophrenia is a common major brain disease whose etiology has not been fully elucidated,and early diagnosis and intervention of schizophrenia are essential.At present,various neuroimaging techniques have become powerful tools to study brain information processing,and scalp electroencephalography(EEG)has become one of the main techniques owing to its unique advantages such as non-invasiveness and high time resolution.Due to the nonlinearity and non-stationarity of EEG signals,it is difficult for traditional EEG analysis methods to capture the complex chaos underlying dynamics of human brain.The brain network has been proven to be an effective frame for understanding the structure and function of the human brain.Existing brain network research mostly focuses on the discovery stage of specific indicators or network connectivity patterns,and how to adjust and control the network states of patients based on these research results remains to be further explored.The typical direction is to promote the transition of abnormal brain networks from one state to another(e.g.,disease state to health state)through global or local perturbations.However,there are still many obstacles on how and where to disturb.From molecule to neural circuit to functional levels,the nervous system has high complexity.Complex network dynamics theory provides a variety of computational models,which help us understand complex nonlinear brain functions and explore the pathogenesis and intervention strategies of brain diseases.Therefore,the dissertation carries out relevant works based on the computational models and combined with the abnormal characteristics of brain network in schizophrenia,focusing on the research of intervention strategies that can promote the change of brain network states.The main innovations and achievements of this dissertation are as follows:(1)The normalized spatial complexity derived indicators were proposed to reveal the abnormal functional connectivity patterns of brain networks in schizophrenia.In view of the problems in the brain network analysis technology that lack of clear quantitative indicators of local functional connectivity and the inability to explore the contribution of various brain regions to the overall functional connectivity level,the dissertation proposed three derived indicators,including global spatial complexity,local spatial complexity of regions of interest and their contribution percentage to the overall functional connectivity level,to reveal the abnormal functional connectivity patterns of brain networks in schizophrenia from the local level.The results found that,compared with the healthy controls,the level of functional connectivity within the frontal lobe decreased,and the level of functional connectivity between the frontal lobe and the parietal occipital lobe increased in schizophrenia patients.Furthermore,from the perspective of contribution to the whole brain,the contribution of the frontal lobe to the whole brain increased in the schizophrenia patients,but decreased in the healthy controls.And the contribution of occipital lobe to the level of functional connectivity of the whole brain was increased,and the level of improvement was significantly lower than that of healthy controls.The dissertation overcomes the limitation of brain network technology only representing from the global level and extends the methodology system of functional connectivity based on graph theory.(2)The brain network reverse evolution control model was proposed to analyze the inherent reasons for abnormal functional connectivity of brain network in schizophrenia.Aiming at the problems in the existing evolutionary algorithms that mostly focus on simulating the forward evolution process,the evolution results poor repeatability rely too much on randomness or accidental,and the number of iterations too much lead to low efficiency of the evolutionary algorithm,the dissertation proposed a brain network reverse evolution control model to simulate the evolution process of schizophrenia patients’ brain functional connectivity towards a healthy direction.And the effectiveness of the reverse evolution control model was evaluated by comparing network similarity and network characteristic parameters between simulated networks and real networks.The results suggest that network synchronization is a characteristic of basal neuron activity in psychiatric disorders and can be used as a specific indicator of brain abnormalities in schizophrenia.And the network topology similarity between the simulated networks and the real networks is significantly improved,and the difference of network topology characteristic parameters disappears after evolution control.Compared with the random or probabilistic increase or deletion edges evolutionary algorithms,this dissertation is more simple,efficient and targeted,providing a new approach to explore the neuropathological mechanism of schizophrenia.(3)The brain network pinning control model was proposed to investigate the specific strategy to drive the brain network to transition from one state to another.Aiming at the issues that network controllability theory cannot guide the actual control algorithm design and control cost,the dissertation proposed a brain network pinning control model,including the criterion for minimum number of pinned nodes,the pinned nodes selection strategy and the pinning control principle,to investigate the specific strategy to drive the brain network to transition from one state to another.And the effectiveness of the pinning control model was evaluated by comparing EEG signals,connection coupling strength,network topology and network synchronization between controlled networks and the driving networks.The results indicated that,the brain networks could theoretically be controlled,but are extremely hard to control through a single brain region.Between the driving networks and the controlled networks,the overall shape and peak distribution of EEG signals are similar,and there is no statistical difference in connection coupling strength(PLVs),and the network similarity increases and network synchronization error decreases.This dissertation offers a specific calculation method for the problems of how and where to disturb in driving the state transition of brain networks,and provides ideas and theoretical methods for the adjustment and control of brain network.(4)The approach of brain network dynamics analysis based on microstates was proposed,to reveal the potential relationship between EEG microstates and brain network dynamics,which is expected to be applied to brain network intervention strategies in the future.In view of the unclear relationship between brain network dynamics and different microstates,whether brain networks occur simultaneously with the life cycle of microstates,and which processes in the brain are associated with overall mutations between different microstates,the dissertation proposed a brain network dynamics analysis approach based on microstates.Moreover,the abnormal electrophysiological characteristics of schizophrenia patients were analyzed from five aspects to reveal the brain network dynamics behind each microstate,including microstate parameters,microstate functional connectivity pattern,microstate synchronization pattern,microstate controllability and microstate synchronization pinning control.The dissertation aims to analyze each state of brain behind the network dynamics and investigates whether specific brain regions and microstate types drive network state changes,or whether all brain regions and microstates contribute equally.The results suggest that abnormal microstate D(fronto-parietal network)may be a biomarker of schizophrenia disease,and microstate B(visual network)may represent a compensatory mechanism that maintains brain function and exchanges information with other brain regions.The topological differences of microstate networks determine their different roles in driving states transition of brain networks.Microstate and functional connectivity provide complementary perspectives on neural dynamics,and the fusion of the two provides potential insights into understanding brain function in health and disease.