A Comparative Study on the Short Time-and Fast-Fourier Based Spectral Analysis of EMG Signals in Dynamic Exercises
CHEN Biao,WANG Zheng-lun,ZHAO Lei,HU Feng-xia,SUN Jing-zhi,HE Li-hua,WANG Sheng,YANG Lei
DOI: https://doi.org/10.13692/j.cnki.gywsyzyb.2012.03.002
2012-01-01
Abstract:Objective With the traditional fast Fourier transform(FFT)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 LT10 and RL3 muscles(paired t-test,P0.05),however,there was no difference if compared in the muscles of RT10 and LL3.The variability of MF by STFT was higher than that by FFT in most cases.There was no difference in the slopes by the two algorithms,whereas the correlation between them was high(r0.5).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(r0.7).Conclusions The traditional FFT can still be used for processing the EMG signals recorded during dynamic exercises,at least at low workload level.Further study is needed to understand the time-frequency analysis in EMG spectrum and its characteristics under different dynamic activities and muscles.