基于外周神经接口的运动功能重建的基础理论与关键技术

The objective of this project is to investigate the neural information encoded in peripheral nerves for controlling reach and grasp functions, develop reliable and adaptive neual decoding algorithms for separating command and sensory signals from the peripheral nerves for the purpose of real-time， peripheral nerve based neural interface controlling a multi-degree multi-finger robotic arm to perform reach and grasp tasks. The key scientific questions to be investigated include 1) coding in the peripherl nerve signals for synergistic functions among muscles involved in different types of grasps, 2) coding mechanism for sensory feedback signls, and 3) adaptive process of the control when learning new grasp skills or during injury recovery for functional improvement..The implementation of the research is to use chronically implanted stable and high performance microelectrode arrays to record neural signals while non-human primates moving their hands to reach and grasp diferent sized, shaped or textured objects at different locations. Recorded data include neural activities from motor and somatosensory cortices, radial, medial and ulnar peripheral nerve bundles, EMGs from major muscles for reach and grasp function,and kinematics of arm and fingers during the movement and grasp. Based on temporal and frequency characteristics of motor and sensory signals, using wavelet and other signal processing techniques to develop algorithms to 1) separate motor and sensory signals from peripheral nerves, 2)decoding motor commands for individual muscle control; 3)decoding motor commands for muscle synergies for grasp types, and 4)decoding sensory signals for position, force and texture information. One innovative approach in this investigation is to perform the recording under various perturbations on grasping object location, size and shape, movement path obstruction, arm freedom restriction, sensory nerve block or topical anasthesia. These interferences on normal movement or neural information transmission will create different task conditions and performance, enhancing or removing sensory influence on feedback control, allowing direct comparison on performance and therefore providing rich informatin to assist neural information decoding algorithm development and validation, enabling reliable and adaptive decoding algorithms for peripheral neural interface to be used on artificial arm control to significantly improve function of amputees or paralyzed individuals. the research project is aimed to deliver a demonstratino of practical implementation of the peripheral neural interface system for real-time control of 3-fingered robotic hand on a 5 degrees of freedom arm.

本项目旨在研究外周神经信号中的控制指令和传感反馈信息，利用神经接口来控制机器人手臂实现伸展抓握任务，从而为截肢或伤残者开发能够进行自主控制的多功能智能假肢。拟解决的关键科学问题包括：外周神经信号的控制指令中协调多组肌肉群执行复杂但灵活的伸展抓握动作的编码；传感反馈信息中各种传感信号的编码；这些编码在受到干扰或病变损伤后的变化规则或规律。研究方法是在实际执行手臂伸展抓握任务时用高性能的神经记录电极实时记录手臂外周神经信号，开发稳定可靠且具有自适应性的神经解码算法分离外周神经束中的下行运动控制指令与上行感觉反馈信息，控制信号指导10自由度三指机械手臂进行实际伸展抓握物体的操作，并结合假肢上的传感器（触觉、力度、运动和位置）信号,通过电刺激外周传感神经将反馈信息输入神经系统，以此构建基于外周神经接口的闭环假肢控制系统。

通过构建外周神经接口可以从外周神经中提取运动和感觉相关的神经信号进而控制机械臂完成伸展抓握任务，从而为截肢或上肢伤残者开发能够进行自主控制的多功能智能假肢。 本项目主要以非人灵长类动物作为实验对象，搭建了非人灵长类植入式神经接口实验平台，主要围绕“外周神经信号记录”，“外周神经信号解析”及“基于外周和皮层神经信号的假肢控制”三个方面开展工作。在外周神经信号记录方面，通过特定参数的定制电极从猕猴前臂外周神经记录信号，利用特定的电极转接口和颅骨固定器，将外周神经电极接口固定在猕猴颅骨上从而实现长期稳定的信号记录。在外周神经信号解析方面，通过分析猕猴上肢伸展抓握过程中对应的外周神经信号和皮层信号，发现外周神经信号的发放模式与行为学任务密切相关，同时通过外周神经信号和皮层神经信号的时频分析，解析了不同行为学任务下皮层神经信号和外周神经信号的联合发放规律。在外周神经信号解析的基础上，联合皮层和外周神经信号，利用支持向量机等分类器实现对伸展抓握模式的离散解码，同时采用卡尔曼滤波等实现基于神经信号的运动轨迹实时预测，进而实现对机械臂伸展抓握三维运动的实时控制。本项目的研究结果具有重大意义，是国内首次实现了皮层神经信号与外周神经信号的同步长期记录，为探究大脑中枢神经系统到外周神经系统的神经信号传递机制提供了有效途径，结合外周神经信号和皮层神经信号的智能假肢控制方法也提供了一种稳定可靠的假肢控制新方法，整个项目推动了神经科学与工程等多个学科的交叉与发展，为基于神经接口的多功能智能假肢自主控制系统的研发提供了理论与技术支持。