基于fMRI脑功能成像的机器人辅助腕手神经康复训练与评价方法研究

It is the injury in the central nervous system that leads to disability after stroke. This project aims to develop a robotic system to work inside the environment of functional brain imaging modality fMRI delivering rehabilitation training to the wrist and hand, and further, to establish a synergetic analysis method for the motion data recorded by the robot as well as the brain activation data by fMRI. This enables exploration of the brain source for limb dyskinesia, and will promisingly help to optimize rehabilitation strategy to achieve optimal recovery. First of all, a spatial impairment model will be proposed to fully characterize the wrist and hand disability. Then, training tasks are designed in accordance with motor learning principles. Inside the fMRI environment, strong electromagnetic fields and limited space bring challenging obstacles for the development of the robot. Actuation and sensing techniques are carefully selected, and a special serial-parallel mechanism is designed. The robot is difficult to control due to time delay, nonlinearity, and strong coupling of force and position. A sliding mode controller is designed, taking into account the feedback as well as feedforward signals by model prediction. Impedance control is implemented to allow smooth switching between free motion and constrained motion of the wrist and hand, and also provides good robustness. Based on synchronized data from the robot and fMRI, correlation of the motion and brain activation is analyzed, which helps to evaluate the efficacy of rehabilitation therapy and eligibility of the assessment criteria. Thus, a systematic methodology has been proposed and validated for the investigation of rehabilitation induced brain re-organization after stroke.

脑卒中致残是一种中枢神经性运动障碍。本项目将研制一个机器人系统在脑功能成像fMRI环境下对腕部手部进行康复训练，并建立一套对机器人动作数据和fMRI神经激活数据的协同分析方法，进而探索运动障碍的神经损伤根源，以期优化康复策略取得最佳康复进展。具体而言，对于腕部和手部，基于运动学习原则设计康复训练任务。在fMRI强电磁环境和受限空间严苛约束下，采用特殊驱动和传感技术，设计机器人机构与腕部手部互动执行康复任务。充分考虑时延、非线性、力和位置强耦合的挑战，结合模型预测前馈和反馈，设计滑模变结构控制器。在此基础上，设计阻抗控制方法，允许局部完全自主运动。基于同步的动作数据和神经激活数据，建立康复训练与脑激活的协同分析方法，进而研究康复训练方法的有效性和评价方法的适用性。由此，本项目为探索康复训练促进脑功能重组的规律，提出了一种行之有效的研究方法。

我国脑卒中防治形势严峻，脑卒中后致残率高。康复训练可以促进神经系统的可塑性，改善患者对肢体行为的控制能力。机器人技术已经成为康复训练的重要方式。在本课题中，我们面向上肢远端灵巧操作的康复需求，建立了一个面向上肢康复的人机灵巧力触觉交互平台，针对康复训练任务设计、人机灵巧力触觉交互方法以及评价方法等展开了深入研究；为了更好的实现康复训练效果向日常生活能力的转移，开展了上肢双侧协调互补ADL运动康复训练，实现了患侧与健侧上肢的差异化人机交互和协作；针对康复训练中的惰性效应，建立了上肢康复训练任务的在线评价与调整方法，基于脑电、肌电、力和运动等多模态信号，对患者的神经参与程度、运动控制能力以及任务完成情况进行多层次在线评价，进而针对性地对任务进行调整；面向fMRI脑功能成像苛刻电磁环境，采用绳牵引串联弹性驱动技术实现对人机交互机构的远程驱动，将控制问题刻画在鲁棒控制架构下，提出有限带宽约束，设计的控制器既保证了人机交互的稳定性，还打破了保守性的局限；针对中枢神经运动区特别是SMA，开展了fMRI研究，探索运动障碍的神经溯源。发表国内外期刊论文9篇、会议论文15篇，其中SCI/EI检索17篇；申请专利3项，其中1项已授权；培养硕士研究生6名，协助培养博士研究生1名；获得国内外学术奖项3项。