The Research Of Feature Extraction Of Stroke EEG

EEG is a new and effective method for the rehabilitation of stroke,and also a hot topic in the research of stroke rehabilitation.The application of motor imagery-brain computer interface(MI-BCI)technology in stroke rehabilitation is a new method for functional rehabilitation of stroke in recent years.Motor imagery is a special way of motor function.It exists in sports memory,follows the principle of central motion control,and also can activate the movement memory but has no obvious movement output.As a method of activating the motor neural network,motor imagery can be used in any stage of the stroke to aid the treatment,in order to achieve the purpose of improving the motor function of stroke patients.But there is still no unified conclusion about the effect of stroke on the neural mechanism of motor imagination.This paper focuses on two characteristics of EEG based on motor imagery cognitive task-brain topographic and brain effect network-to reveal the effect of stroke on the motor imagination neural mechanism.In view of the research of brain topographic map,this paper mainly did two aspects of work.On the one hand,extracted the brain topographic of the stroke patients group and the control group at each stage of the motor imagery cognitive process,and then proposed a classification method based on the vector space angle method of the brain topographic map.On the other hand,obtained the group averaged brain topographic map of two groups of subjects at each stage.The results showed that,the reduction and loss of the active degree of the patients with lateral parietal lobe and the movement related area of the stroke side in the stage of mental rotation of visual stimuli and response cognitive process,which resulted in the decrease of the spatial cognitive ability of stroke patients,and the performance for brain activity was ipsilesional hypo-activation in the response cognitive process.In the research of brain effect network,a dynamic causal model based on adaptive directional transfer function was proposed,the brain effect network of the stage of Visual stimuli encoding was constructed.Then studied the differences between stroke group and control group according to the properties of brain network.The results showed that,the effect connection strength of stroke patients reduced because of the influence of the lesion of stroke.And during visual stimuli encoding,the cortical activity pattern of control group is the activation of occipital lobe,frontal lobe and parietal lobe.But stroke patients showed that Hypo-activation in frontal and left parietal lobe,and right temporal lobe activation.The mentioned two EEG features were deeply researched in this paper.The brain activity during each sub-stage in motor imagery could be observed intuitively by brain topographic.For measuring brain activity,this paper did the research from a brain network view.Effective connectivity calculated by dynamic causal interactions could reflect the information flow among different brain regions.The method was new relatively.Both two kinds of tasks have a certain practical value for the research of stroke rehabilitation and the establishment of the stroke warning system in the later stage.