基于表面肌电多尺度熵的递增负荷诱导肌肉疲劳评估

Assessing Muscle Fatigue Induced by Increasing Load Exercise Based on Multiscale Entropy Analysis of Surface Electromyography Signals

目的:探讨通过表面肌电(sEMG)信号的多尺度熵分析评估运动负荷变化时肌肉动态疲劳等级的可行性。方法:10名男性大学生完成从50~200+W的1~5级功率自行车递增负荷蹬车试验,每级负荷运动采集最后5 s的股直肌、股内侧肌和股外侧肌的sEMG信号进行多尺度熵计算,根据样本熵随尺度1~30的变化曲线获得平均多尺度熵指标用于评价肌肉疲劳。结果:sEMG信号的多尺度熵曲线随负荷递增而整体下移,其中负荷1与负荷2、负荷4与负荷5的多尺度熵接近,分别反映肌肉未疲劳和轻度疲劳、以及中度和重度疲劳状况;各级负荷下多尺度熵随尺度增加先快后慢,但尺度较小时相邻负荷数值接近,其中尺度为1(即传统样本熵)时重叠交错最为严重;当尺度大于10时多尺度熵曲线逐渐平缓并拉开间隔,从上往下按负荷递增顺序排列。结论:sEMG信号的多尺度熵更准确的表示了肌肉系统的复杂度,相比传统样本熵能更客观的反映肌肉的工作状况和疲劳等级;多尺度熵分析不仅计算量小,而且能适应时变负荷下肌肉动态收缩的复杂性,实际工作中可将归一化的平均多尺度熵作为衡量肌肉动态疲劳程度的一个量化指标。

Objective：This paper presents a multiscale entropy（MSE）based complexity measure of surface electromyography（sEMG）signals recorded from RF,VM and VL muscles for assessing increasing load exercise induced muscle fatigue during cyclic dynamic contraction. Methods：Ten male undergraduate stu-dents were selected to perform cycle ergometer experiment with 1 to 5 step increasing load from 50 watts to 200 watts. The last 5 seconds sEMG signals on eachload were acquired for MSE analysis,The mean of multiscale entropy（MMSE）was derived from MSE curves from scale 1 to scale 30 and then was used to accessthe increasing load induced muscle fatigue. Results：MSE curves declined with load increment as a whole but load 1 and load 2,load 4 and load 5 approachescloser which represent the case of fatigue-free and severe fatigue respectively;The MSE values increased with scale factor and curves between neighboring loadscan be well separated over time scale 10 at which the curves approach parallel and distribute from up to down in order with load level. However the MSE of neigh-boring loads overlaps for small scales especially at scale 1,namely conditional sample entropy. Conclusions：The mean value of MSE over all scales representsthe complexity level of skeletal muscle activity so that it can be used to evaluate muscle＇s working statues and fatigue degree with more accuracy than condition-al sample entropy. The MSE method is computationally easy compared to time-frequency analysis meanwhile it is more adaptable to complicated conditions dur-ing various dynamic contractions.It is suggested the normalized mean of multiscale entropy（NMMSE）be employed as an index to assess muscle fatigue inducedby dynamic varied-loading tasks.