基于非负矩阵分解和复杂网络的肌间耦合分析

Intermuscular Coupling Analysis Based on Non-negative Matrix Factorization and Complex Networks

目的利用表面肌电信号,探究多块肌肉之间的信息传递关系和不同握力下的肌肉之间的耦合特性。方法采集6名健康受试者在不同握力下的表面肌电信号,首先用广义偏定向相干计算多通道肌肉之间的相干性;然后用非负矩阵分解算法将相干性进行分解;最后用复杂网络建立不同条件下的肌肉功能网络,利用图论的方法,定量分析肌肉功能网络的连接特性。结果不同握力下肌肉的激活程度存在显著性差异;肌间耦合在10～20Hz频段上较为显著,且在10～20Hz频段上耦合程度随着握力的增加呈现显著性变化。结论肌间耦合在不同频段和不同握力下呈现显著性不同,表明了中枢神经系统维持不同握力的控制模式,本文方法为诊断运动功能障碍和评价康复效果提供了依据。

Objective To explore the relationship of information transmission between multiple muscles and the intermuscular coupling characteristics under different grip strengths with surface electromyography(sEMG).Methods Six healthy subjects were asked to perform grip under different strengths,and sEMG were collected from 7 muscles.Firstly,the coherence between multi-channel muscles was calculated by Generalized Partial Directed Coherence.Then the coherence was decomposed by NMF.Finally,CN was used to establish the muscle function network under different conditions.Results It was found that there was a significant difference in the activation degree of muscles under different grip strengths;the intermuscular coupling was more significant in the 10~20 Hz band;and the coupling degree in the 10~20 Hz band showed a significant change with the increase of the grip strength.ConclusionIntermuscular coupling is significantly different in different frequency bands and different grip strengths,indicating that the central nervous system maintains different grip strength control modes,which provides a basis for the diagnosis and rehabilitation evaluation of motor dysfunction.