面向脑卒中运动功能评估的多模态神经-肌肉功能耦合分析及机制研究

Reasonable and effective function assessment for stroke is of great significance to the formulation of rehabilitation strategies. The widely accepted assessment technique for the clinic mainly focused on the subjective and qualitative evaluation of the observation indicators from the rehabilitation physicians. However, it is lacking of effective analysis and quantitative evaluation of multimodal EMG-EEG information interaction and motor mechanism in the motor control system. For this, the aim of this study is to explore the method of multimodal neuro-muscular coupling (MNMC) analysis based on EEG and EMG signals from the perspective of multimodal neuromuscular function coupling. Firstly, a novel time synchronization and association analysis models will be established to explore the dynamic interaction and network patterns in the cerebral cortex. Secondly, the frequency-band-based interaction and time-delay interaction index for bi-direction will be proposed to describe the functional interaction and time-delay characteristics for neuro-muscle system. Then, the cross-frequency coupling and time-frequency synergy analysis method will be studied to analyze the muscle activation level and the synergistic mode. Combined with the clinical test and experimental research, the association characteristics and information interaction mechanism of MNMC features in healthy people and stroke patients will be excavated to analyze the evolution of MNMC in different lesions and stages of rehabilitation. After that, the multimodal neuro-muscular coupling characteristic indicators associated with clinical scales will be screened out. Additionally, a qualitative-quantitative assessment strategy based on the neuro-muscular coupling and clinical scale will be obtained, which provides a feasible basis for the effective assessment and rehabilitation treatment for stroke.

脑卒中运动功能的合理有效评估对康复策略制定具有重要意义。临床常用评估手段主要依据医师对外在观测指标的主观定性评价，缺少对内在运动控制系统的多模态脑肌电信息交互及运行机制的有效分析和量化评估。为此，本项目从多模态神经-肌肉功能耦合分析角度，基于头皮脑电和表面肌电信号开展多模态神经-肌肉功能耦合分析方法研究：构建时序同步和定向关联分析模型探究大脑动态连接及网络模式，建立频带耦合及双向时延交互指标描述大脑与肢体间功能耦合及时延特性，研究时频协同和异频耦合方法分析肌肉激活水平和协同模式；进一步结合临床测试和实验研究，分析健康及脑卒中患者的多模态功能耦合特征间关联特性和信息交互机制，探究不同病灶和康复阶段的多模态功能耦合演变规律，筛选与临床定性评价指标相关联的功能耦合特征指标，构建基于多模态功能耦合和临床评定量表的定性-定量综合评估策略，为脑卒中运动功能状态有效评估和康复治疗策略制定提供可行依据。

本项目围绕脑卒中运动功能状态的智能化量化评估问题，从神经肌肉系统的多模态神经-肌肉耦合(MNMC)层面，拓展了现有脑电和肌电信号的同步分析，从局部频带、时间延迟、多尺度等特性层面，分析脑卒中患者后大脑与肌肉间信息交互特性，为揭示脑卒中运动功能障碍的病理机制提供依据。具体内容包括：在功能脑区耦合振荡研究方面，分别提出了多尺度多变量传递熵和大尺度功能网络模型分析方法，用于探究功能关联脑区间的多尺度及大尺度耦合特性，并指出脑卒中患者不同脑区间的尺度特性消退；在皮层肌肉功能耦合研究方面，分别提出局部多尺度传递熵、时延信息传递指数、复杂多尺度传递熵等方法，用于分析大脑与肢体肌肉间的局部频带、时间延迟及尺度特性分析，并指出脑卒中患者beta频段功能耦合降低甚至缺失、延迟时间增加、异频功能耦合减弱；在肌肉协同及耦合研究方面，分别提出变模态分解-一致性、小波包-非负矩阵分解等方法，用于分析不同肌肉间时频协同及耦合特性，并指出脑卒中患者肌肉协同模块数目、动作完成程度、变异性和肌肉激活度均发生变化。在此基础上，本项目基于皮尔逊相关法分析了多层次神经-肌肉功能耦合特征指标与临床量表间的关系，并筛选出局部频带、异频耦合及多尺度多个耦合特征指标与临床量表具有显著关联关系；进一步，构建了基于雷达图方法的运动功能可视化评估模型，实现对脑卒中患者运动功能的多元可视化定性-定量综合评估；并将评估后的结果进行综合评价函数值的计算，判定评价结果的有效性。这为脑卒中运动功能状态有效评估和康复治疗策略制定提供可行依据。