基于模式识别的肌电控制用于脑卒中上肢康复的研究

The most common disability following stroke is a motor function dysfunction such as hemiparesis. Specifically, upper-limb function is often limited in the contralateral side of a brain lesion, and restoration of the upper-limb function is known to be difficult, which diminishes health-related quality of life. This proposal is designed to conduct a comprehensive exploration on the solutions to several key problems regarding the application of pattern-recognition-based myoelectrical control on the upper limb rehabilitation for stroke survivors. One aspect of our efforts will be made for the study on myoelectric pattern recognition, for the purpose to propose novel and effective methods for processing surface electromyographic (EMG) signals measured from paretic muscles of a patient’s upper limb and identifying the patient’s various movements or movement intentions under specific conditions of muscle weakness and involuntary spasticity resulted from central nervous system injury (i.e., stroke), therefore establishing a theoretical foundation for rehabilitation training using multiple degrees-of-freedom control. The other aspect of our efforts will be made for the analysis of abnormal movement patterns of the hemiparetic upper limb, in order to investigate the underlying mechanisms of upper-limb movement disorders in term of neuromuscular control property, and to quantitatively assess the motor functions of the paretic upper limb, thus providing guidelines to the design of customized rehabilitation plan for different stroke survivors. Moreover, clinical evaluation experiments will be performed to assess the actual effect of the proposed methods on upper limb rehabilitation after stroke. The research conducted in this proposal helps to better understand physiological pathological mechanisms of neuromuscular control after stroke, and is of great importance in terms of seeking for solutions to the restoration of upper-limb functions and hand dexterity in particular, and thus improving stroke rehabilitation therapy.

脑卒中后运动功能障碍发生率较高，且偏瘫上肢康复困难，严重影响患者的生活质量。本课题旨在深入探索基于模式识别的肌电控制应用于脑卒中上肢康复的若干关键问题。一方面，开展肌电模式识别方法研究，针对在神经损伤条件下存在的肌力减退和非随意挛缩现象，提出能有效处理偏瘫上肢肌电信号并识别多种上肢精细运动或运动意图的新方法，为实现多自由度控制的康复训练奠定理论基础。另一方面，开展偏瘫上肢异常运动模式分析研究，从神经肌肉控制层面探索不同患者上肢运动失调的内在机制，定量评估上肢的运动功能，从而为个性化的康复计划设计提供指导。此外，开展临床实验以验证所提出的方法应用于上肢康复训练的效果。本课题的研究成果将有助于更好地理解神经肌肉控制机理和脑卒中后偏瘫病理，对解决偏瘫上肢尤其是手部康复的难题，提高脑卒中康复治疗水平具有重要的意义。

本课题采用高密度表面肌电电极阵列、肌电模式分析与识别、异常运动模式分析和神经肌肉病变检测等先进的表面肌电采集与处理技术，以脑卒中后偏瘫上肢为研究对象，对患侧上肢肌电模式识别与神经信息解码、偏瘫上肢的异常运动模式分析、偏瘫肌肉病变检测，以及基于模式识别的肌电控制用于脑卒中上肢康复效果验证等问题开展了深入研究，取得了以下重要研究成果：（1）为解决在神经通路损伤条件下运动控制信息解码的难题，在肌电采集、预处理、特征提取、分类、控制策略等方面提出了一系列新型肌电模式识别方法，形成了自然多自由度肌电控制方法框架；（2）发挥可穿戴的表面肌电和惯性传感器数据融合进行运动捕获优势，提出偏瘫上肢运动功能和肌痉挛程度的量化评估方法；（3）提出新型肌电干扰相分析方法，实现对偏瘫肌肉病变无创检测，并发现偏瘫肌肉相比健侧和健康肌肉发生和发展有复杂多样的神经肌肉病变；（4）设计了手部外骨骼机器人，并对肌电控制驱动外骨骼机器人的康复治疗效果进行临床验证。本课题的研究成果不仅为偏瘫上肢尤其是手部康复提供了有效的治疗技术、提高偏瘫康复治疗水平，而且有助于揭示脑卒中后偏瘫的病理，指导制定更加科学高效的治疗策略，对脑卒中康复具有重要临床应用前景。