In this work, in order to restore the movements of lower limbs for paralyzed patients, a functional electrical stimulation (FES) rehabilitation system based on brain-computer interface (BCI) was developed. BCI adopts the technology of steady-state visual evoked potential (SSVEP) based EEG. Linear discriminated analysis was used as the classifier to process frequency-domain features of EEG. Five types of motion intention can be recognized, I. E. Start, fast, slow, stop and idle states. The intention recognized was translated as a command to trigger FES system, and stimulated the concerned muscles of lower limbs to generate movements, meanwhile joint angles were measured. Six healthy subjects took part in the pure BCI experiments, and two of the subjects took part in experiments of swinging and walking movements controlled by BCI-FES separately. During swinging experiment, rectus femoris was stimulated. During walking experiment, iliopsoas, glutaeus maximus, rectus femoris, and hamstrings of two legs were stimulated. FES artifacts on EEG were analyzed during experiments. The results showed that the motion intention was well recognized ( average recognition accuracy rate was over 85% ), and the subjects can realize the expected lower limb movements via FES accordingly.%研发基于脑机接口控制的功能性电刺激系统,服务于下肢的运动康复.脑机接口采用的是基于稳态视觉诱发电位的脑电信号技术.使用线性判别分析分类器来处理脑电信号的频域特征,实现对下肢5种运动状态的控制意图识别,即开始、快速、慢速、停止和空闲状态.识别的意图转化为指令触发电刺激系统,刺激下肢的相关肌肉产生运动,并测量关节角度.设计的系统在6位正常受试者上进行了单纯的脑机接口实验,其中2位分别进行了脑机接口控制的小腿摆动与下肢行走的电刺激实验.在小腿摆动实验中刺激的是股直肌,行走运动实验中刺激的是两条腿的髂腰肌、臀大肌、股直肌和腘绳肌.实验分析了电刺激尾迹对脑电信号的影响,结果表明设计的脑机接口可以准确地识别运动意图(平均识别率高于85％),并能够实现电刺激作用下与该意图相对应的下肢期望运动.
展开▼
