Study On Motion Intention Recognition From The Integration Of EEG And Joint Angles For Active Rehabilitation Training System

Some patients of stroke,or spinal cord injury,and traumatic brain injury have different degrees of motor dysfunction in lower extremity.Most of the traditional physiotherapy techniques adopt passive training.However,brain-computer interface based on electroencephalography(EEG)signals,combined with functional electrical stimulation(FES),can integrate active and passive training to improve the effect of rehabilitation training.In this paper,the spontaneous EEG signals during motor imagery(MI)are under study to extract motion intention,in combination with the angles of the elbow joint.Then an active rehabilitation training system is established,which controls FES to stimulate the tibialis anterior of lower extremity to achieve dorsiflexion of the ankle joint,and provides the basis for the lower extremity rehabilitation training.The main research content includes the following aspects:(1)Natural interaction is needed while most EEG systems depend on evoked potentials.An experimental paradigm based on motor imagery EEG signals is designed.The spontaneous EEG signals are collected from multiple subjects,and the brain switch motion intention similar to "switch" is extracted from the motor imagery EEG signals for discrimination.The offline average recognition rate is 90%,and the online average recognition rate is 78%,and send a brain switch control commands every 5.5 s.(2)The factors such as inattention and fatigue during motor imagery tasks affect EEG signals,and some subjects may have the "BCI-illiteracy" problem.Attention level is explored from electrooculography(EOG)signals,and the correlation between attention and motion intention recognition is studied.Then by detecting the attention,the offline recognition rate of motor imagery EEG signals can be improved.(3)From spontaneous potentials only few types of identifiable tasks and is hard to be applied in EEG systems.An experimental paradigm collecting multi-source biological information is designed by including joint angles.The joint angles and EEG signals are integrated at the decision level.A hybrid control strategy is proposed to implement active and passive training.Its upper structure is the secondary control strategy and the lower structtre is the shared control strategy.(4)While most of the traditional rehabilitation training systems adopt passive training,an active rehabilitation training system for lower extremity is established.The system is modular in design,including the master module and the FES module-Subjects control the FES to stimulate the anterior tibial by their motion intention.With the intensity of FES increased,the angles of ankle joint show more obvious dorsiflexion and electromyographic(EMG)signals of the anterior tibial are increased.Additionally,the connection between the sensory motor area of the brain is strengthened after the stimulation of FES.In this paper,motion intention was extracted from EEG signal and gesture signal respectively for research,and an active rehabilitation training system integrating EEG signal and gesture signal control was designed and established.It can provide a research basis for active rehabilitation treatment in the future.