多模式人体刺激与生机电仿人假肢手的增强式生机接口研究

One of the main reasons why the current prosthetic hand products can not reproduce the dexterous functions of human hand is the poor transmission performance of bio-mechanical interface. The electromyographic bio-mechanical interface based on pattern classification method can achieve the multi-mode motion decoding accuracy of more than 95% in intact limbs, however, it has not been adopted in the practical applications because of time-shift characteristics of the physiological signals and EMG signal source missing or degradation of physically disabled patients. The project is proposed to study the experimental paradigms, which are associated with multi-mode stimuli, limb motion decoding and neurophysiology, and the stimulus encoding methods to explore the external stimuli based enhanced bio-mechancial interface technology for dexterous prosthetic hand and its integration mode. The research work will be focused on the following issues with the theoretical analysis and experiments: 1) the time-shift characteristics of the associated features between EMG signals and hand joint movement, and the adaptive decoding methods, 2) the space-time coding and control methods of external stimuli, including electrical stimulation, vibration stimulation, visual stimulation, and etc., 3) the evolvement dynamics of neurophysiological characteristics under stimulations, 4) the experimental paradigms associated with multi-mode stimuli, hand joint movement, and neurophysiology, and the development of their implementation systems. The goal of this project is to propose the neurophysiological feature reconstruction methods with external stimuli, which will take advantage of the plasticity of nervous system, and to provide scientific basis for development of the enhanced bio-mechanical interface with high transmission rate and the dexterous prosthetic hands.

目前的产品化假肢手之所以无法复现人手的灵巧功能最主要的原因之一是生物/机械接口的传输性能差。基于模式分类方法的肌电生机接口虽然能在健全肢体上实现多模式动作95%以上的解码准确率，但由于生理信号特征的时移特性及肢残患者肌电信号源缺失/退化等问题，无法进入实际应用。项目拟通过研究多模式刺激／肢体运动解析／神经电生理关联实验范式及刺激编码方法，探索基于外部刺激的灵巧假肢增强式生机接口技术及其集成方法：从理论分析和实验两方面入手，研究肌电信号与手关节运动关联特征的时移特性与自适应解码方法，外部刺激（电刺激、振动刺激、视觉刺激等）的时空编码与控制方法，刺激作用下神经电生理特征变化的动力学规律，多模式刺激与手关节运动、神经电生理关联实验范式与实现系统构建等问题。目标是利用神经系统的可塑性，提出外部刺激辅助的神经电生理模式特征重建方法，为高传输率、增强式生机接口及灵巧假肢研制积累科学基础。

项目以灵巧假肢的生机接口为研究对象，重点开展肌电信号与手关节运动关联特征时移特性与自适应解码方法、多模式刺激的时空编码与控制方法两大关键科学技术的研究，以解决生机接口应用的鲁棒性性差的问题，从而为高传输率、增强式生机接口及灵巧假肢的研制提供技术与方法的支持。.项目结合脑电研究了刺激诱发的神经响应特性，设计面向生机接口自增强的分布式机械/电刺激方案。研究了基于经颅电刺激（tDCS）的自适应特征增强方法，提高了各动作模式的可分性，显著降低残肢肌电信号分类的错误，同时明显缩短长期截肢患者肌电接口的训练时间。开展了假肢肌电接口长时服役情况下受试者生理变化对肌电特征影响的研究，发现了该变化与肌电接口解码性能的关联规律；重点开展了自适应鲁棒肌电接口的研究，针对肌电电极偏移/肌肉疲劳/残肢肌电源空间受限等因素，提出了自适应肌电解码方法。提出了肌电信号特征平稳子空间投影的共模成分分析法，建立了残肢肌电信号的无漂移特征模板，在保证多模式运动解码率的同时大幅度提高了接口的稳定性。在无需再训练的前提下，可实现假肢长期使用过程中6种动作模式的稳定控制。该方法已集成到灵巧假肢产品中，实现了接口的“近零”再训练。搭建了生机接口的测试与训练系统，设计了可穿戴式多通道肌电采集设备和多信息混合式生物信号传感器；结合康复工程需求，设计并改进了灵巧假肢样机（SJT-6S），可佩戴于截肢患者残端，满足截肢患者80%日常生活的需要；设计并完成了仿生触觉传感器、电刺激器和虚拟手系统，可用于增强生机接口，提高患者残端肌肉的自主控制能力。.课题研究成果在JNE、IEEE TSNRE、IEEE TBME等重要学术期刊和国际会议发表论文36篇，其中SCI收录论文21篇。新增发明专利申请13项，其中授权9项。研究成果获国家技术发明二等奖1项。