Research On Brain Network Patterns Of Motor Imagery Based On Magnetic Resonance Imaging

Motor imagery(MI)is a multidimensional high-level cognitive ability.It is normally defined as the mental rehearsal of actions from a first-person perspective,without any overt physical movement.Many previous studies have reported that MI is beneficial to motor skill learning,stroke patient rehabilitation and brain computer interface(BCI)control.In spite of these benefits,there exist debates over the rationale behind the application of MI.Hence,the further identification of the neural mechanism underlying MI will facilitate the clarification of its related functional roles and provide the necessary neural basis for clinical and engineering applications.This dissertation mainly focuses on the exploration of the neural mechanism underlying MI by using the brain network analysis to investigate the specific network patterns of MI at various levels.In the current work,the patterns of fronto-parietal attention network(FPAN)during resting state,the time-varying network patterns of activated regions during left-and right-hand MI tasks,and the large-scale network patterns of functional networks during resting-state and the two MI tasks have been deeply probed.Moreover,based on these functional connectivity strengths,we predicts the individual’s MI-BCI performance and behaviours by using a multivariate pattern analysis.This paper mainly includes three parts:1.Research on the relationships between the functional and structural patterns of FPAN and MI.Based on the functional and structural MRI data,we evaluated the structural and functional features of FPAN.We found that the eigenvector centrality(EC)and cortical thickness(CT)of the left inferior parietal lobule(IPL)were negatively correlated with MI performance,the CT of the right dorsolateral prefrontal cortex(DLPFC)was negatively correlated with MI performance,while the degree centrality(DC)of right ventral intraparietal sulcus(v IPS)was positively correlated with MI performance.These findings reveal that the individuals who have an efficient FPAN will perform better on MI.In addition,we also found that the EC and CT of the left IPL could effectively predict the individual’s MI performance with 83.3% accuracy.2.Research on the patterns of time-varying networks during left-/right-hand MI.Based on the MI tasks f MRI data,we newly used adaptive directed transfer function(ADTF)and graph theory to investigate the dynamic network patterns of MI.First,we used the generalized linear model to evaluate the activated regions for left-and right hand MI.Then,the regions of interest were defined by using the activated regions.Finally,the time-varying network was constructed based on the activated regions.Our findings showed that the left anterior insula plays an important role in modulating information during the initial stage of MI.The time-varying networks displayed a lateralization during the latter stage of MI,such as the center of network from another regions switched to left SMA during right-hand MI,while the center of network from another regions switched to right SMA during left-hand MI.Moreover,we calculated the network properties of time-varying networks for each subject across each state.We found that the network efficiency was relatively low during the initial stage of MI.Then,the network efficiency gradually increases and keeps stable.Our findings facilitate to better understand the neural mechanisms of MI from the patterns of time-varying networks.3.Research on the patterns of large-scale network during left-/right-hand MI and resting-state.We constructed large-scale network by using the task and resting-state fMRI data and group spatial independent component analysis,and calculated the corresponding event related desynchronization(ERD)for two diverse MI tasks by using MI-EEG data.Our results showed that(a)the task-related functional brain networks are systematically engaged during MI,where specifically the somato-motor network(SMN)and dorsal attention network(DAN)play critical roles in discriminating MI context-specific;(b)an effective interaction among networks could facilitate MI performance;and(c)the connectivity strengths of within-and between-network could serve as a biomarker to predict individual’s behaviors with 81.82% accuracy by using support vector machine.These findings contribute to the understanding of the underlying neural mechanisms of MI from large-scale network patterns.