Developing a brain-machine interface for an ankle robot

描述（由申请人提供）： 该提案建立在使用非侵入性脑电描记术（EEG）捕获可用于控制计算机和/或设备的大脑激活模式的最新进展的基础上。随着技术先进的机器人矫形器的共同发展，在身体康复中使用这种脑机（BCI）或脑机接口（BMI）的潜力现在是一种真实的可能性。为下肢开发这样的BMI设备可以为恢复退伍军人和其他中风和其他神经损伤以及下肢截肢后残疾的人的移动功能和生活质量提供新的途径。在巴尔的摩弗吉尼亚州的马里兰州运动和机器人卓越中心，我们已经开发了一个阻抗控制踝关节机器人（anklebot），以提高瘫痪踝关节运动控制和步态功能的中风幸存者。最近，我们还首次使用非侵入性（EEG）解码和重建非残疾受试者的跑步机行走下肢关节运动学。我们的研究结果是非常相似的，从灵长类动物的直接（侵入性）皮层录音，因为他们进行双足踏车行走。踝关节机器人提供了一个理想的测试平台，可以整合这两条研究路线，为下肢开发创新的BMI。在这项提案中，我们将验证一种新的神经解码方法，用于提取踝关节运动的运动学参数，同时使用高密度EEG记录在皮质下中风后慢性轻偏瘫患者及其年龄匹配的对照组。我们还将确定解码参数的重测信度。两组受试者将在佩戴踝关节机器人的同时接受训练，以生成大脑激活模式来自主控制踝关节机器人，证明下肢动力矫形器可以通过使用皮层信号来操作的概念验证。据我们所知，这将是下肢BMI应用的首次演示。具体而言，该试点项目旨在测试高密度EEG可以可靠地解码与踝关节机器人中进行的背跖屈（DF，PF）运动相关的不同大脑激活模式的假设。基于我们先前使用相同方法来训练上肢的BCI控制的经验，我们期望从EEG信号重建踝运动将产生足够高的相关性和信噪比，以允许从解码的EEG进行闭环学习，使得控制组和中风组将学习生成用于控制踝关节机器人移动通过预期DF的特定EEG模式。PF运动这项研究将提供第一个概念证明，“智能”下肢假肢可以通过使用非侵入性脑电图的意图控制。这些结果将证明我们的长期目标的可行性，开发一个思想控制的外骨骼双足运动。

DESCRIPTION (provided by applicant): This proposal builds on recent advances in using noninvasive electroencephalograhy (EEG) to capture brain activation patterns that can be used to control computers and/or devices. The potential to use such brain-computer (BCI) or brain-machine interfaces (BMI) in physical rehabilitation is now a real possibility with the co-evolution of technologically sophisticated robotic orthoses. Developing such BMI devices for the lower extremites could provide new avenues for restoring mobility function and quality of life among Veterans and others who are disabled after stroke and other neurological injuries, as well as lower limb amputation. In the Baltimore VA's Maryland Exercise and Robotics Center of Excellence, we have developed an impedance controlled ankle robot (anklebot) to enhance paretic ankle motor control and gait function in stroke survivors. Recently we also have shown the first use of noninvasive (EEG) to decode and reconstruct lower extremity joint kinematics for treadmill walking in nondisabled subjects. Our results are very similar to those obtained with direct (invasive) cortical recordings from primates as they performed bipedal treadmill walking. The anklebot provides an ideal test platform to integrate these two lines of research to deveop an innovative BMI for the lower extremity. In this proposal we will validate a novel neural decoding approach for extracting kinematic parameters for ankle movements recorded with the anklebot while using high-density EEG in persons with chronic hemiparesis after subcortical stroke and their age-matched controls. We will also determine the test-retest reliability of the decoded parameters. Both groups of subjects will be trained while wearing the anklebot to generate brain activation patterns to autonomously control the anklebot, demonstrating proof-of-concept that a lower extremity powered orthoses can be operated by using cortical signals. To our knowledge this will be the first demonstration of a lower extremity BMI application. Specifically, this pilot project is designed to test the hypothesis that high density EEG can reliably decode distinct brain activation patterns associated with dorsi-plantarflexion (DF, PF) movements performed in the anklebot. Based on our prior experience using the same approach to train BCI control for the upper extremity, we expect that reconstruction of ankle movements from the EEG signals will yield correlations and signal-to-noise ratios sufficiently high to permit closed loop learning from the decoded EEG, such that the control group and the stroke group will learn to generate specific EEG patterns for controlling the anklebot to move through intended DF-PF movements. This study will provide the first proof of concept that "smart" prostheses for the lower extremity can be controlled by intention using noninvasive EEG. These results will demonstrate feasibility for our long-range aim of developing a thought-controlled exoskeleton for bipedal locomotion.