Electrocorticographic studies of human cortical function

描述（由申请人提供）：我们的总体目标是在两年内展示DARPA革命性上肢假肢的人类闭环控制，基于皮质电图（ECoG）信号的实时解码。在视觉反馈下，我们的人类受试者将实现对假肢足够的皮质控制，以伸出，抓住和操纵真实物体。我们的合作团队将在模块化假肢（MPL，由JHU- APL开发）的神经控制算法开发和测试方面积累丰富的经验，这源于darpa赞助的RP2009项目。基于我们独特的专业知识和经验，我们准备用一种创新的方法，在人类受试者和动物身上进行平行实验，来迎接RFA的大挑战。在我们开发和测试基于脑电图的神经控制算法的同时，为了癫痫手术的临床目的，在动物身上进行的平行研究将提供更一致和长期的实验时间来验证控制算法，并允许对电极放置和配置进行更深入的研究，包括研究皮层内记录的峰值和局部场电位（LFPs）以及表面记录的脑电图之间的关系。我们将测试一个假设，即利用不同时间和频率尺度的ECoG频谱特征，可以实现对假肢的开环和闭环控制，例如，低频用于慢速和/或粗动作，高频（bbb70 Hz）用于快速和/或个性化动作。基于这些特征，受试者将使用假肢在3D空间中执行中心伸出任务，协调抓取需要不同手形的物体，包括手腕和5个手指的动作。脑控将在目标肢体运动的闭环视觉反馈下，通过ECoG的实时信号处理和假肢的驱动来实现。闭环控制将首先使用假肢的虚拟现实模型进行演示，随后使用JHU-APL模块化假肢本身。结果将通过成功率、试验完成时间、轨迹/抓取形状与本地肢体运动的相似性以及整体学习/适应率来评估。

DESCRIPTION (provided by applicant): Our overall goal is to demonstrate within two years human closed-loop control of the DARPA Revolutionizing Upper Limb Prosthesis, based on real-time decoding of electrocorticographic (ECoG) signals. Under visual feedback, our human subjects will achieve sufficient cortical control of the prosthesis to reach out, grasp, and manipulate real objects. Our collaborative team will build on its substantial experience in developing and testing neural control algorithms for the Modular Prosthetic Limb (MPL, developed by JHU- APL), which arose from participation in the DARPA-sponsored RP2009 program. Based on our unique combination of expertise and experience, we are poised to meet the RFA's Grand Challenge with an innovative approach using parallel experiments in human subjects and in animals. While we develop and test ECoG- based neural control algorithms in patients implanted for the clinical aims of epilepsy surgery, parallel studies in animals will afford more consistent and long-term experimental time to validate the control algorithms, and allow deeper investigation into electrode placement and configuration, including investigating the relationship between intra-cortically recorded spikes and local field potentials (LFPs) and surface-recorded ECoG. We will test the hypothesis that both open- and closed-loop control of the prosthetic limb can be achieved using ECoG spectral features at different time and frequency scales, e.g. low frequencies for slow and/or coarse movements and high frequencies (> 70 Hz) for rapid and/or individuated movements. Based on these features, subjects will use the prosthetic limb to perform a center-out reach task in 3D space with coordinated grasping of objects requiring different hand conformations that incorporate both wrist and 5 finger actions. Brain control will be implemented with real-time signal processing of ECoG and actuation of the prosthetic limb under closed-loop visual feedback of object-targeted limb movements. Closed-loop control will be first demonstrated using a virtual reality model of the prosthetic limb and subsequently using the JHU-APL modular prosthetic limb itself. Outcomes will be assessed with measures of success rate, time to trial completion, trajectory/grasp-shape similarity to native limb movements, and overall learning/adaptation rate. PUBLIC HEALTH RELEVANCE: Project Narrative This project will demonstrate the feasibility of using the signals recorded from non-penetrating electrodes on the surface of the brain to allow patients who have lost arm and/or hand function to intuitively control a revolutionary new prosthetic limb with far greater versatility and life-like dexterity than previously available prostheses. This could have a profound long-term impact on the ability of future generations of patients seeking to restore lost upper limb function with a lifelike prosthesis.