Pj. Sparto et al., Wavelet analysis of electromyography for back muscle fatigue detection during isokinetic constant-torque exertions, SPINE, 24(17), 1999, pp. 1791-1798
Study Design. An investigation of the effects of human trunk extensor muscl
e fatigue on the temporal change in frequency content of the electromyogram
as quantified using the Fourier and wavelet transforms during the performa
nce of repetitive dynamic trunk extension.
Objectives. To evaluate whether alterations in the Fourier and wavelet tran
sform measures were consistent with a shift of the signal power to lower fr
equencies, and to determine which measures were more highly correlated with
the decline in maximal trunk extension torque.
Summary of Background Data. Objective assessment of trunk muscle fatigue is
likely to play a more important role in the rehabilitation and prevention
of low back injuries, given the association between lack of trunk muscle en
durance and acquisition of low back pain. Validation of new methods designe
d to quantify the level of fatigue using the surface electromyogram is nece
ssary before these techniques can be used in industrial rehabilitation sett
ings. The wavelet transform is a recent development in the signal processin
g of electromyograms that shows promise as a method for assessment of fatig
ue.
Methods. Trunk muscle electromyograms obtained from study participants perf
orming repetitive isokinetic trunk extension endurance tests were analyzed
using the wavelet and the traditional Fourier methods. Trunk extension torq
ue was controlled at 35% and 70% of the participants' maximal voluntary con
traction while they exerted at 5 and 10 repetitions per minute. The decline
in maximal trunk extension torque was measured once per minute. Linear reg
ression quantified the rate of change in Fourier and wavelet measures cause
d by fatigue, whereas Pearson's correlation coefficient determined their as
sociation with the decline in maximum torque.
Results. Changes in the characteristics of the electromyogram were consiste
nt with a shift to lower frequencies: The signal power at higher frequencie
s was reduced, whereas the power at lower frequencies was elevated. The amo
unt of change was dependent on the task conditions (exertion level and repe
tition rate). The wavelet-based measures demonstrated as strong an associat
ion with the decline in maximal torque output as the Fourier-based measures
.
Conclusions. This study demonstrates:that:, assessment of trunk muscle fati
gue during isokinetic movementis possible using both Fourier and wavelet me
asures. However, the methods were not as likely to change significantly dur
ing lower rates of exertion. These methods, when implemented in a controlle
d setting, may be used to document the rehabilitation process and guide pre
ventive exercise training.