Study Of Motor Imagery Cognitive Process For Stroke Patients Based On Electroencephalography

Motor impairment after stroke is a major cause of permanent disability. Loss of entirefunction or partial ability of movement brings lots of inconvenience to the stroke patientsand their families. Recovery of motor function in order to improve life quality of patients isa primary aim after stroke. Motor imagery was recently reported to be useful for recoveryof motor function in stroke patients. Motor imagery is a cognitive process in which therepresentation of a specific motor action is internally reactivated within working memorybut without an overt motor output. Motor imagery could be used as an additional trainingmethod during any stage after stroke to active movement network and it’s not dependent onpatients’ residual physical function and patients could participate actively. Neuroimagingstudies showed that cortical neural networks including posterior parietal and visual cortex,premotor cortex (BA6), supplementary motor areas (SMA) and primary motor cortex (M1)were activated during motor imagery. In addition, some subcortical structures, e.g., basalganglia, were also activated during motor imagery. Basal ganglia and motor cortices havebeen known to be involved in motor planning and execution, and their activation duringmotor imagery suggested that actual and mentally simulated movements largely share thesimilar neural structures. Compared with passive movement rehabilitation, motory imageryis more in line with the normal excitation mode of neural system, which is from centralneural system (e.g., brain structures) to extremes (e.g., limb), and more effective toregenerate the reflex neural arc of movement.Although motor imagery training has justified some optimism in stroke rehabilitation,the majority of these studies to date were poorly controlled cases or with small samplesizes and the outcomes were inconsistent. Furthermore, the underlying neural mechanismsof motor rehabilitation by motor imagery have not been fully understood. The studies sofar have been mostly based on behavior assessment and overall neural activations byneuroimaging technology during motor imagery with few temporal details of the cognitiveprocess. Therefore, one interesting question is how stroke lesion would affect the neural mechanism of each sub-stage during the cognitive process of motor imagery. In this thesis,we aim to use event-related EEG recordings with high temporal resolution and behaviorperformance to find answers for those scientific questions:(1) stroke lesion effect onoverall behavior performance during motor imagery;(2) stroke impact on phase-lockedcortical activation in each sub-stage during motor imagery;(3) stroke impact onphase-unlocked cortical activation in each sub-stage during motor imagery;(4) strokeimpact on the interdependence between different brain areas and network properties fromfunctional segregation and integration perspective. We hope this study could reveal moreneural cognitive mechanism of motor imagery during rehabilitation after stroke which iscrucial for treatment planning and clinical outcome quantitative estimation.In this thesis, a quantitative motor imagery cognitive process was employed, i.e.,identification task of hands pictures which were rotated to different spatial angles (mentalrotation task, MRT). Event-related electroencephalography (EEG) signal with hightimporal resolution was used to reveal temporal-spatial patterns of neural activation for twogroups in different cognitive sub-stages during MRT. Behavior performance, event-relatedpotential (ERP), event-related (de)synchronization (ERD/ERS), phase synchronization andcognitive functional network were investigated to illustrate neural mechanism of strokeeffect on cognitive process of motor imagery. Specific results were described as follow:(1) In order to evaluate the overall impact of stroke on motor imagery, behaviorperformance of stroke patients was assessed. Behavior results showed that stroke patientswho with better clinical outcomes had better performance in MRT, which indicatedbehavior performance during MRT had good correlation with clinical outcomes in strokepatients. Patients with stroke lesion in left hemisphere could also accomplish MRT, showedsignificant “angle effect”. Although movement impairment of affected (right) hand wasobserved, representation of the affected hand movement was still intact or partly preserved,i.e., patients could accomplish the MRT with affected hand. Compared with controlsubjects, longer response time and lower accuracy rate were observed for stroke patients,which indicated that stroke lesion affected patients’ behavior performance in MRT.(2) Stroke impact on phase-locked activation of cortex during each sub-stage in motorimagery was assessed by event-related potential (ERP). Two significant ERP components(i.e., P200and P300) were observed during MRT, which were indicated visual stimuliencoding and mental rotation sub-stages respectively. During visual stimuli encodingsub-stage, stroke patients showed only frontal and parietal lobe in contralesionalhemisphere activated. Hypo-activation in bilateral central area and frontal lobe inipsilisonal hemisphere of stroke patients indicated that only physical property of handpictures were encoded and no higher movement related cognitive process were involved for stroke patients during visual stimuli encoding sub-stage. Significant “amplitudemodulation effect” in control subjects during mental rotation of visual stimuli wasobserved in bilateral parietal lobes which played an important role in spatial cognition.Stroke patients showed significant contralesional parietal lobe activation (i.e., lateralizationeffect), but no “amplitude modulation effect” was observed. All ERP results indicated thatstroke patients showed significant deficiency of movement-related areas activation inipsilesional hemisphere.(3) In addition, event-related (de)synchronization (ERD/ERS) in beta band wasevaluated to assess the stroke impact on cortical phase-unlocked activation in eachcognitive sub-stage during motor imagery. Stroke patients had significantly smallerbeta-ERD values than control subjects. During visual stimuli encoding, only occipital andparietal lobes activated in stroke patients. Hypo-activation in movement related brain areasfor stroke patients might contribute to impairment of visual stimuli encoding ability.During mental rotation of visual stimuli, bilateral frontal, central and parietal lobes showedbeta-ERD for control subjects while stroke patients showed smaller beta-ERD valuesparticularly in frontal lobes. During response cognitive process, significant “angle effect”of beta-ERD was observed for control subjects, more activation particularly in righthemisphere was observed for more difficult task. No such effect was observed for strokepatients and the phase-unlocked activation was localized in parietal, central and occipitalareas. Hypo-activation in movement related brain areas in stroke patients might contributeto poor ability of spatial information encoding and impairment of movement executionability during MRT.(4) Besides local phase-locked and phase–unlocked cortical activation was evaluated,interdependances between different brain cortical areas were assessed by phasesynchronization to investigate the stroke effect on cortical interactions. Furthermore, neuralfunctional network baesd on graphy theory was constructed to reveal cognitive networkalterations after stroke. Stroke patients showed significantly smaller phase synchronizationindex in ipsilesional hemisphere and interhemisphere. Stroke lesion might contribute to thesynchronization reduction in brain areas around lesion in the left hemisphere andsynchronization between brain areas with long distance in two hemispheres. Functionalnetwork of stroke patients also showed “small-worldness” property while with longercharacteristic path length and smaller clustering coefficient which indicated more pathswere needed to transfer information and the local interaction was sparse. All these resultsindicated that ability of functional segregation and functional integration after stroke wasimpaired significantly. Statistical results showed that patients showed significant largernodal clustering coefficient and betweenness in right (contralesional) hemisphere, particularly in mental rotation sub-stage, which showed compensation effect ofcontralesional hemisphere. These results revealed that functional network after strokereflected a lower capacity to integrate the communication between distance brain regionsand lower tendency to be modular. Patients’s network also showed contralesionalhemisphere compsenssation effect.In summary, we investigated the cognitive alternations during MRT after the lefthemispheric stroke injury using both behavior and electrophysiological methods.Movement impairment after stroke would also accompanied by poor behaviorperformances in motor imagery. Hemispheric lateralization after stroke was observedwhich was due to the ipsilesional hypo-activation during visual stimuli perception andmental rotation. Furthermore, stroke lesion also resulted in the loss of “angle effect” due tothe impaired spatial information processing during response execution. Synchronization inipsilesional hemisphere was impaired significantly and the network of stroke patientsshowed lower ability to integrate the communication between brain areas and lowertendency to be modular, and contralesional hemisphere compensation effect. These resultsmight provide new insights into the understanding of cognitive process during mentalrotation, and could be referenced as a guide in stroke rehabilitation with motor imagerytraining. These were the major work in this thesis. On one hand, these studies investigatedstroke effect on cognitive process in motor imagery with time course from bothphase-locked and phase-unlocked event-related oscillations. On the other hand, strokeeffect on functional segregation and integration of the cognitive functional network wasalso investigated by graphy theory based on phase synchronization. This thesis would becrucial to reveal cognitive neural mechanism of motor imagery after stroke and potentiallyuseful for clinical application in rehabilitation training after stroke.