| Objectives:Disorders of consciousness(DOC)is a state of loss of consciousness caused by various severe brain injuries.The rapid development of network science provides us with valuable tools to analyze the human brain.The resting-state brain is composed of several discrete networks,which remain stable for 10–100 Ms.These functional microstates are considered the building blocks of spontaneous consciousness.In recent years,the development and improvement of the source-reconstruction technology have made it possible to study the temporal dynamics of the brain with high temporal and spatial resolution at the source level of electroencephalogram(EEG).And the rapid development of network science provides us with valuable tools to analyze the human brain.Characterizing brain networks can not only improve our understanding of the mechanisms underlying brain functions such as perception and attention,but also lead to the discovery of "neural markers" related to the onset or progression of neurological diseases.However,most studies assumed static functional connectivity between distinct brain regions during the scan period,but little is known about possible changes in the temporal dynamics of dynamic functional connectivity(d FC)in DOC patients.To address this question,we examined different time-scale spatiotemporal dynamics of EEG oscillation amplitudes recorded in different consciousness state.And the spatiotemporal dynamics were assessed across multiple frequency bands.We sought to explore the relationship between states of consciousness and spatio-temporal dynamics at different time scales,looking for possible biomarkers of consciousness.Next,We aimed to analyze microstates in the resting-state EEG source space in patients with DOC,the relationship between state-specific features and consciousness levels,and the corresponding patterns of microstates and functional networks.Finally,to evaluate the effect of long-term high-density transcranial direct current stimulation(HD-t DCS)on multi-dimensional brain network indexes of EEG by tracking the changes in consciousness level of patients.Methods:1.In chapter 3,we obtained resting-state EEG data from 90 patients with DOC(46with minimally conscious state [MCS] and 44 with vegetative state [VS]/unresponsive wakefulness syndrome [UWS]).We used the sliding windows approach to construct d FC matrices.Then these matrices were clustered into distinct states using the k-means clustering algorithm.We conducted a state analysis of the resting-state EEG.These include number of transitions(NT),frequency(F),mean duration time(MDT),and transition syntax.Next,we compared the differences in state properties between the 2groups of patients with different levels of consciousness at different time scales and frequency bands and used multivariate pattern analysis(MVPA)to identify meaningful time scale frequency bands.Finally,we build a decoding model based on support vector machine(SVM)algorithm to predict the state of consciousness.2.In chapter 4,a total of 84 DOC patients were included in this study.Coma Recovery Scale-Revised(CRS-R)was used to evaluate the level of consciousness of patients.Of these,27 patients were in MCS and 57 were in VS.The resting state EEG data of the patients were collected and preprocessed to obtain clean and usable data.eloreta method was used for source reconstruction of the data.A microstate analysis of EEG time series in source space was performed.The rank sum test was used to compare the differences in microstate attributes between the two groups.Finally,MVPA method was applied to explore the correspondence between microstates and specific brain networks.3.In chapter 5,42 patients with DOC were enrolled,and their consciousness level was evaluated by CRS-R scale,including 13 VS patients and 29 MCS patients.We recorded the CRS-R scores of the patients before and after stimulation.The patients were divided into responder group(R+)and non-responder group(R-).Then,the EEG network indexes,spectral power,and normalized spatial complexity(NSC)of the two groups were statistically analyzed.Results:1.In chapter 3,by the elbow rule,we confirm that the optimal number of clustering states is four.MVPA suggested that the dynamic functional connectivity of fast time scale β band,medium time scale β band and γ band could distinguish the level of consciousness.Compared with the MCS group,patients with VS had decreased dynamic functional connectivity across the forebrain range.Temporal properties,flat MDT,and NT were also significantly different between VS patients and MCS at different time scales in the high frequency band.By using multi-band and multi-time scale d FC as features,the prediction model is constructed based on SVM algorithm,and its classification accuracy is 83.3%,and the transition syntax feature weight is the highest.2.In chapter 4,according to the elbow rule,the optimal number of microstates is seven.Each microstate has a different spatial distribution of cortical activation and contains different brain networks.MVPA revealed that different functional connectivity patterns were associated with source-level microstates.There were significant differences in the microstate properties,including spatial activation patterns,temporal dynamics,state shifts,and connectivity construction,between the MCS and VS groups.3.In chapter 5,CRS-R scores improved significantly after 14 days of stimulation as compared to baseline.R+ had increased spectral energy in several electrode channels in the alpha2 band and beta1 band,predominantly in the frontal and parietal electrodes.In the global metrics of graph theory,the global efficiency,local efficiency,and smallworld property sigma were higher in the R+ group in the alpha1 band than before,and the small-world property sigma in the beta1 band higher than before.In contrast,for Rgroup there was no difference before and after stimulation.In theta band,local efficiency decreased after stimulation for the R+ group,no difference was seen in global efficiency and small-world property sigma,and for the R-group,global efficiency increased compared to before.Similarly,in terms of node properties.the Cz and Pz node clustering coefficients(NCC)increased in the alpha1 band for the R+ group compared to the pre-stimulation,while no difference was observed for the R-group.In terms of NSC,the R+ group increased in the beta2 band,while R-increased in the alpha1 band.Furthermore,the prediction model was developed using NSC,and its accuracy is 0.939,yielding the highest spatial complexity weights in the alpha band and gamma band.Conclusion:1.Loss of consciousness is accompanied by imbalances in complex dynamics within the brain.Meanwhile,the brain operates in different patterns at different time scales and frequency bands.The spatio-temporal dynamics of the brain,rather than the state of the brain itself,play a more important role in the generation and recovery of consciousness.Among them,the transition of frontal(frontotemporal parietal)and occipital states in high frequency bands(β and γ bands)is more important in the discrimination of consciousness states.These findings may contribute to a better understanding of abnormal brain networks in DOC patients and the search for possible biomarkers as targets for neuromodulation.2.Our findings suggest that consciousness depends on complex dynamics within the brain and may originate from the anterior cortex.In VS patients,cortical activation was mainly concentrated in the posterior occipital region,with loss of connection with the forebrain.The limitation of consciousness to the posterior cortex may account for its low level of consciousness.The study helps elucidate the markers that are specific and generalizable during consciousness generation and recovery,as well as the neurophysiological mechanisms of human consciousness.3.The improvement in the level of consciousness after stimulation in DOCs is reflected to varying degrees on multidimensional metrics of the EEG.Selective modulation of the resting network can be observed on the EEG.It is possible that the recovery of a stronger integration in the alpha1 band and a stronger isolation in the beta2 band will be crucial in the recovery of consciousness.Also,modulation of the posterior parietal lobe can lead to an EEG response related to consciousness in DOC,and the posterior cortex may be one of the key brain areas involved in the formation or maintenance of consciousness. |
PDF Full Text Request