Research On The Plasticity Of Brain Network In Motor Imagery

The application of motor imagery(MI)in the neurological recovery of patients with motor dysfunction(e.g.,stroke,spinal cord injury)has become a hot topic of research in recent years.Recently,the approaches of brain network analysis have been broadly applied in MI-based research.Studying the changes in brain network plasticity associated with motor dysfunction during MI can help us further understand the brain functional network mechanisms of motor disorders.In this thesis,the brain network analysis combined with graph theory is used to construct electroencephalography(EEG)functional networks of stroke and spinal cord injury patients.The mechanisms of their brain plasticity have been analyzed and explored.The specific work in this thesis is as follows:1.Research on classification and recognition of MI based on brain network fusion features.MI intention recognition is an important part of studying the plasticity of brain network in the procedure of MI.In this study,we have calculated the spatial characteristics of the coherence adjacency matrix,the common spatial pattern and the fusion characteristics of patients with spinal cord injury and stroke respectively.A support vector machine is used for classification and recognition.The experimental findings reveal that the fusion features have got good performance in the classification experiments of the two groups of subjects.Among them,the classification of stroke patients based on fusion features has the highest accuracy rate of 88.2%.2.Research on brain functional network of stroke patients based on the procedure of MI.In this thesis,the adaptive directed transfer function is used to construct the time-varying network of patients with stroke and healthy control group,to study the dynamic brain network mechanism in the procedure of MI,and finally to study the potential relationship between global efficiency and Fugl-Meyer scale through correlation analysis.The results of the study have concluded that there is no significant change from laterality to bilateral symmetry in the brain network of stroke patients during the procedure of MI,mainly in the primary motor area contralateral to the brain injury and in the frontal lobe with enhanced effective connectivity to nonmotor areas.And the connectivity between the occipital lobe and non-motor areas is enhanced.There is a significant correlation between Fugl-Meyer scale and global efficiency.The foregoing results will help us better understand the dynamic brain network mechanism of stroke patients during MI.3.Research on the reorganization network of patients with spinal cord injury in the procedure of MI.This thesis has used the method of coherence to construct the functional network of patients with spinal cord injury at different stages of MI,and statistically analyzes the network properties.Finally,a Coherence-Resnet Graph Convolutional Networks(C-Res GCN)algorithm is proposed.The results have shown that brain connectivity is weaker in the α and β bands during the MI state than in the resting state.In the γ-band,we have found that patients with spinal cord injury show enhanced connectivity in the primary motor area and a weakened default mode network in the frontoparietal lobe during the MI.The C-Res GCN model proposed in this thesis gets good performance with the highest classification accuracy of 96.25%,and the average accuracy of γ-band is higher than the average classification accuracy of α and β.This study helps to a deeper understanding of the mechanisms of brain function after spinal cord injury as well as providing new insights into MI-based BCI systems.The study of brain network plasticity can help us to understand more deeply the brain function mechanism of patients with related diseases,and also provide a new reference for the rehabilitation treatment of related diseases.