Research on the real-time control system of lower-limb gait movement based on motor imagery and central pattern generator

Abstract Achieving one-to-one correspondence between the subjective mind and limb movement can further improve the effect of rehabilitation. This paper proposes a novel control scheme wherein the ideal physical rehabilitation training—‘what you think is what you move’—is performed by decoding the mind. First, for feature extraction and pattern classification, the common spatial pattern (CSP) algorithm and the linear discriminant analysis (LDA) algorithm were applied to electroencephalogram (EEG) signals collected by OpenBCI. Second, the result of motor imagery (MI) recognition was transferred from OpenViBE to Arduino UNO through the virtual reality peripheral network (VRPN). The spinal cord stimulator received corresponding commands to generate pulse signals of different polarities. Finally, an electrode pair was used to stimulate specific central pattern generator (CPG) site on the spinal cord of rats for rehabilitation training. Eight subjects participated in the classification experiment (highest accuracy rate, 91.38%). This system realised the gait-like movements corresponding to the subject's mind through functional electrical stimulation (FES). The subject with the best recognition result was selected and participated in the gait rehabilitation training experiment of rats with spinal cord injury (SCI). The Basso, Beattie, and Bresnahan (BBB) scale score and grip strength were significantly better after three weeks in the SCI + electrical stimulation (ES) group than in the SCI group. The results proved the effectiveness of the novel system for motor function recovery. This study provides new ideas applicable to the existing methods of motor function rehabilitation and has immense scientific research value in rehabilitation medicine.