短时和快速傅里叶变换用于动态活动下表面肌电频谱分析的比较研究  

A comparative study on the short time-and fast-Fourier based spectral analysis of EMG signals in dynamic exercises

作　　者：陈飚[1] 王正伦[1] 赵磊[1] 胡凤霞[1] 孙敬智[1] 何丽华[2] 王生[2] 杨磊[1] 

机构地区：[1]华中科技大学同济医学院公共卫生学院,教育部环境与健康重点实验室,湖北武汉430030 [2]北京大学公共卫生学院,北京100083

出　　处：《工业卫生与职业病》2012年第3期144-150,共7页Industrial Health and Occupational Diseases

基　　金：十一五国家科技支撑计划项目(2006BAI06B08)

摘　　要：目的短时傅里叶变换(short time Fourier transform,STFT)作为频谱分析的新方法比快速傅里叶变换(fast Fourier transform,FFT)更适合于非稳定信号,探讨这两种方法在分析评价动态活动肌肉疲劳的差异。方法9名男性大学生志愿者在实验室从事模拟的搬举活动(变换搬举重量和距离),测定搬举时左右两侧胸10和腰3水平竖脊肌的表面肌电信号。另10名志愿者在功率车上以3种速度跑步,记录此间双腿股内侧肌和股外侧肌的肌电。分别用STFT和FFT求得活动期间每秒钟肌电信号的中位频率(median frequency,MF),计算MF值依时间变化回归的斜率,比较两种算法所得结果的差异。结果搬举试验,STFT所得MF值多小于FFT,左侧胸10(LT10)、右侧腰3(RL3)竖脊肌比较,差异有统计学意义(P<0.05);右侧胸10(RT10)、左侧腰3(LL3)比较,差异无统计学意义(P>0.05)。STFT变异性多大于FFT。两种MF斜率比较,差异无统计学意义(P>0.05),然而二者间呈高相关性(r>0.5)。STFT与FFT疲劳检出率比较,差异无统计学意义(P>0.05)。跑步试验,两种方法计算所得4块肌肉MF比较,差异均无统计学意义(P>0.05)。STFT变异性多大于FFT。MF斜率和疲劳检出率比较,差异均无统计学意义(P>0.05)。两种MF斜率仍然呈高相关性(r>0.7)。结论至少在低负荷情况下,FFT依然可用于动态活动下肌电信号的频谱分析。仍值得进一步研究这两种频率变换方法以及EMG信号在不同活动和肌肉的特性。Objective With the traditional fast Fourier transform （EFT）as reference, a new spectrum processing method of short time Fourier transform （STFT） was used for analyzing the EMG signals from different muscles during lifting tasks and running on treadmill, to compare the differences, if any, by the two methods in evaluating muscle fatigue during dynamic activities. Methods 9 student volunteers were engaged in 6 groups of simulative lifting activities, i. e. , the lifting weight and changing distance between the groups. The EMG signals were recorded simultaneously from the muscle erector spinae left and right at thorax 10 （LT10,RT10）, and lumbar 3（LL3,RL3）. Another 10 student volunteers were involved in running on a treadmill with 3 different velocities. The EMG signals were recorded simultaneously from the vastus medialis and vastus laterals at left and right legs. EMG signals were processed using the both the methods of STFT and FFT. Median frequency（ MF） was calculated from the recorded signals for every 1 minute. The slopes of MF were calculated by regression as an indicator of muscle fatigue. Results In the lifting experiment, STFT --based MF was generally lower than that by FFT, with significant differences in I,T10 and RL3 muscles（paired t -- test, P ~ 0.05）, however, there was no difference if compared in the muscles of RT10 and I.I.3. The variability of MF by STFT was higher than that byFFT in most cases. There was no difference in the slopes by the two algorithms, whereas the correlation between them was high（r^0.5）. There was also no significant difference in the fatigue detection rate by two methods. In the running exercises, there was no difference in the MF of four muscles calculated by two methods. The variability of STFT--based MF was still higher than that with FFT. No difference was found either in slopes or in fatigue detection rates. Again, the slopes by two methods were correlated with each other（r〉0.7）. Conclusions The traditional FFT can still be used for proce

关 键 词：表面肌电图 快速傅里叶变换 短时傅里叶变换 动态活动 肌肉疲劳 

分 类 号：R131[医药卫生—劳动卫生]