中文摘要：

目的 本文对小儿脑瘫患者行走过程中下肢主要肌肉表面肌电信号进行基于多变量AR模型的相干性分析,以探索脑性瘫痪对下肢肌肉活动的影响.方法 利用多变量自回归模型的方法计算正常组和两组不同级别脑瘫组儿童在行走过程中下肢胫前肌、腓肠肌、股直肌和股外侧肌两丽之间的相干系数进行统计学分析,比较不同组之间的差异性特征.结果 股外侧肌与股直肌信号相干系数在3组之间具有显著性差异（P＜0.05）,其中Ⅰ-Ⅱ级脑瘫患者相干系数高于正常组,Ⅲ-Ⅳ级脑瘫患者低于正常组,而其他肌肉间的相干性在不同组之间无显著性差异（P＞0.05）.结论 相干性系数可作为一种探索肌肉活动相互间关系的手段,从神经控制的角度为研究小儿脑瘫患者异常步态的形成原因提供参考.从相干性分析中还发现中枢神经系统对肌肉活动的控制存在差异性.

英文摘要：

Objective Multivariate autoregressive model based on EMG--EMG coherence is used to analyze the lower extremity muscle activities for children with cerebral palsy （CP） when walking. Methods EMG signals were collected from four muscles： tibialis anterior （TA） , gastrocnemius （GA） , rectus femoris （RF） and vastus lateralis （VL） of each leg for both CP groups and typical developed （TD） group, and pairwise muscle coherence values were calculated and compared among the three groups. Results Among six pairwise muscle coherence, only VL-RF displayed significant difference among the groups （P 〈 0.05）. Ⅰ-Ⅱ level CP group had higher coherence and Ⅲ-Ⅳ level CP group had lower coherence than TD group. Conclusions EMG-EMG coherence provides a method to study muscle connections during walking, especially for the children with CP. From coherence analysis we find that central nervous system in the control of muscles is different.