杨基海 杨洪宁 周 平 章劲松 孙路遥 (中国科学技术大学，合肥 230026) 中国生物医学工程学报 1999 0 18 1


    关键词：EMG；自动分解；迭加的运动单位动作电位波形 期刊 zgswyxgcxb 0 生物医学信息 fur -->


    

本文提出一种不需人工干预的EMG信号自动分解算法。算法的学习阶段使用初始的一小段EMG记录以估计不同运动单位的动作电位模板。分解阶段根据运动单位发放统计特性，采有基于最小错误率的贝叶斯准则，将EMG信号分解为构成它的各运动单位动作电位序列，这里重点讨论了具有迭加动作电位波形的EMG信号分解问题。最后利用合成的模拟EMG信号及真实EMG信号对算法进行了检验，表明其具有很高的分解正确率.

    分类号： R318.04; R540.41

STUDY ON AUTOMATICDECOMPOSITION OF EMG SIGNALS

    WITH OVERLAPPING MUAP WAVEFORM

Yang Jihai, Yang Hongning, Zhou Ping, Zhang Jinsong, Sun Luyao

(University of Science and Technology ofChina, Hefei, Anhui, 230026)

ABSTRACT A scheme of EMG signaldecomposition was reported no need of human supervision. A learning stage was firstapplied at the beginning of recording to estimate typical spike shapes of different motorunits (MUs). As for the decomposition stage, a method was devoloped, with specialconsideration of temporal spike overlap. It minimized the probability of error, takinginto accout the statistical properties of discharges of MUs. The method was tested on areal recording as well as synthetic data.

     ......