[论文解读] Brain-Computer Interface Controlled Robotic Gait Orthosis: A Case Report

An H., Po T. Wang+3 more|arXiv (Cornell University)|Aug 24, 2012

EEG and Brain-Computer Interfaces参考文献 14被引用 4

一句话总结

本研究展示了基于脑-计算机接口（BCI）的机器人步态矫形器（RoGO）在脊髓损伤（SCI）患者中恢复行走能力的可行性。通过基于脑电图（EEG）的运动想象，离线模型达到了94.8%的准确率，线上测试显示每场次仅0.8次误报，且与神经意图的交叉相关性达0.809，表明神经意图与机器人运动之间具有高度同步性。

ABSTRACT

Reliance on wheelchairs in individuals with spinal cord injury (SCI) leads to many medical co-morbidities. Treatment of these conditions contributes to the majority of SCI health care costs. Restoring able-body-like ambulation after SCI can potentially reduce the incidence of these conditions, as well as increase independence and quality of life. However, no biomedical solution exists that can reverse this loss of neurological function, and hence novel methods are needed. Brain-computer interface (BCI) controlled lower extremity prosthesis may constitute one such novel approach. An able-bodied subject underwent electroencephalogram (EEG) recording while engaged in alternating epochs of idling and walking kinesthetic motor imagery (KMI). These data were analyzed to generate an EEG prediction model for online BCI operation. A commercial robotic gait orthosis (RoGO) system (treadmill suspended), was interfaced with the BCI computer. In an online test, the subject was tasked to ambulate using the BCI-RoGO system when prompted by computerized cues. The performance of this system was assessed with cross-correlation analysis, and omission and false alarm rates. The offline accuracy of the EEG prediction model was 94.8 +/- 0.8. The cross-correlation between instructional cues and the subject's BCI-RoGO walking epochs averaged over 5 online sessions was 0.809 +/- 0.056 (p-value<10^-5). There were on average 0.8 false alarms per session and no omissions. These results provide early evidence that restoring brain-controlled ambulation is feasible. Future work will test this system in individuals with SCI. If successful, this may justify development of BCI-controlled lower extremity prostheses for free overground walking for those with complete motor SCI. This system may also be applied to incomplete motor SCI to improve neurological outcomes beyond those of standard physiotherapy.

研究动机与目标

• 开发一种脑-计算机接口（BCI）系统，使脊髓损伤（SCI）患者能够通过该系统控制机器人步态矫形器（RoGO）实现行走。
• 评估利用脑电图（EEG）信号在运动觉运动想象（KMI）过程中实现脑控行走的可行性。
• 通过线上运行中的交叉相关性、遗漏率和误报率评估系统性能。
• 为未来在完全运动性SCI患者中实现BCI-RoGO系统的临床转化提供基础。

提出的方法

• 在一名健全受试者执行静息与行走运动觉运动想象（KMI）交替阶段时，记录其脑电信号。
• 利用记录的数据离线训练EEG预测模型，以从脑电信号中分类行走意图。
• 将BCI系统与商用跑步机悬吊式机器人步态矫形器（RoGO）连接，实现实时控制。
• 线上测试通过计算机化提示触发行走，系统性能通过提示与BCI输出之间的交叉相关性进行评估。
• 性能指标包括遗漏率（未检测到的指令）和误报率（错误激活）。
• 使用p值 < 10^-5评估交叉相关性结果的统计显著性。

实验结果

研究问题

• RQ1 基于EEG的运动想象能否可靠预测机器人步态矫形器的实时行走意图？
• RQ2 BCI-RoGO系统在神经指令与机器人运动同步性方面的表现如何？
• RQ3 在线BCI-RoGO运行中，遗漏率与误报率如何比较？
• RQ4 系统是否能在受控的线上环境中实现高准确率与低误报率？
• RQ5 该系统的性能是否足以支持未来在完全运动性SCI患者中的临床应用？

主要发现

• 离线EEG预测模型准确率达到94.8 ± 0.8%，表明从EEG信号中对行走意图的分类能力很强。
• 指令提示与受试者BCI-RoGO行走阶段之间的交叉相关性为0.809 ± 0.056，p值 < 10^-5，证实了显著的同步性。
• 系统每场次平均仅记录0.8次误报，表明非预期激活的发生率极低。
• 在五次线上测试中未记录到任何遗漏，表明对预期行走指令的检测极为可靠。
• 结果为在受控环境下使用BCI-RoGO系统实现脑控行走提供了早期证据。
• 研究结果支持未来开发BCI控制的下肢假体，以实现完全运动性SCI患者在自由行走环境中的步行功能。