.I. Electromyogr.

Kinesiol. Vol. 5, No. 4, pp. 239-243, 1995

Copyright 0 1996 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
1050.6411(94)0006-9 105~6411195 $10.00 + 0.00

Muscle Fatigue and Calibration of EMG Measurements

Tommy oberg
Department of Biomechanics and Orthopaedic Technology, University College of Health Sciences, Jiinkiiping, Sweden

Summary: Amplitude electromyography (EMG) is often used asan estimator

of muscular load. Such measurementscan, however, be biased by other
factors, for example muscularfatigue. The aim of this study was to examine
the influence of fatigue on amplitude parameters of the EMG. The test
subjectsraised the arm to 90 of abduction in the plane of the scapula.The
hand was loaded with 0, 1 and 2 kg during 5, 3 and 2 min respectively. EMG
wasrecordedfrom the trapeziusmuscle,and spectraland amplitude parameters
were calculated. There was a significant rise of the EMG amplitude as a sign
of fatigue at all load levels: 7% min- at 0 kg, 15% min- at 1 kg, and
19% min- at 2 kg. At 0 kg hand load there was no change of the spectral
parameters but at higher load levels there was a significant decline of mean
power frequency: 3% min- at 1 kg and 11% at 2 kg. The amplitude rise due
to muscle fatigue may seriously jeopardize calibration measurementsunless
the duration of the load is kept limited.

Key Words: Calibration-EMG-Fatigue-Amplitude-Frequency.

J. Electromyogr. Kinesiol., Vol. 5, 239-243. December

INTRODUCTION and fatigue effects. One of the classical fatigue signs

Measurements of the electromyographic (EMG)
amplitude have been widely used to estimate
muscular contraction levels but there are a number
of pitfalls in such estimations. There is a quantitative
relationship between the EMG signal and the
contraction level but this relationship is often non-
linear. Furthermore, the measurements may be
influenced by difficulties in establishing the degree
of contraction (because of difficulties in defining
the maximal voluntary contraction (MVC) for a
single muscle), noise, filtering characteristics of the
skin, changed firing rate of motor units, recruitment
ReceivedMay 10, lY94.
RevisedSeptember 28. 1994.
Accepted September 28. 1994.
Addresscorrespondence and reprint requests to T. Oberg,
Director/Associate Professor, Department of Biomechanics and
Orthopaedic Technology. University College of Health Sciences,
P.O. Box 1038. S-551 11 Jiinkaping, Sweden,
in the EMG is a rise of the amplitude, despite
constant external load2,. Possibly it may reflect an
increased internal load or lowered efficiency of
the muscle. Consequently fatigue effects must be
considered if amplitude parameters are used to
calibrate the EMG.
In surface EMG, the amplitude is relative and it
must be related to some kind of reference contraction
or calibration contraction,s,6. Generally such a
calibration contraction is performed at the beginning
of the measurement but H5gg et al. have proposed
test contractions at regular intervals throughout the
recording session. Not even the most thorough
calibration procedure could, however, compensate
for many of the problems related to amplitude, for
example the fatigue effects. For this reason it may
be important to include other indicators of fatigue
in the measurements.
Ever since Kogi and Hakamada (1960)9 described

239
240 T. &ERG

a spectral shift towards lower frequencies in sustained EMG Recording and Analysis
static contractions, spectral parameters derived from
the EMG in the frequency domain have been used Surface EMG was recorded from the trapezius
as such indicators of muscle fatigue2,3. The most muscle at a point half way between the spinous
frequent parameters are mean power frequency process of C7 and the acromion. Pre-gelled surface
(MPF), median frequency (MDF), and the number electrodes (Medicotest child ECG electrodes,
of zero crossings (ZC) 8,10. Problems related to spectral Olstycke, Denmark) were used. They were placed
parameters as estimators of muscle fatigue have been close together with an interelectrode distance of
discussed by Gberg et a1.-14. 22 mm. The diameter of the gel cushion was 10 mm
The aim of this study was to quantify fatigue and the metal surface beneath it 4 x 4 mm. The
effects on amplitude and spectral parameters of the signal was preamplified (10x) and amplified with
EMG and to discuss the use of spectral parameters equipment built in our department. The EMG signal
to detect fatigue in the amplitude EMG. was stored on a FM tape recorder (Teat R-71,
Tokyo, Japan). It was later replayed through a
fourth order Butterworth anti-aliasing filter with an
MATERIALS AND METHODS upper cut-off frequency of 800 Hz, AD converted
with a sample rate of 2 kHz, and analysed on
Subjects and Test Procedure a Digital Equipment PDP 11/44 computer. The
frequency content was calculated by fast Fourier
Eleven subjects, all women, with a mean age transform (FFT). A quality control routine was
of 37.6 (21-55) yr were examined. During the applied to the signal 1,6. The signal was analysed in
examination they were sitting on an ordinary office sequences of 0.5 s. Root mean square (rms) and
chair. The straight arm was elevated to 90 in the mean power frequency (mpf) were calculated.
scapular plane, with 0 kg hand load during 5 min
(Figure 1). Four subjects also performed the same
procedure with 1 kg and 2 kg hand load during 3 Statistical Analysis
and 2 min respectively. The subjects were first
tested with 0 kg, then 1 kg and 2 kg. There was a The 10-s data were further analysed on a personal
rest period of 5-10 min between the tests. computer with a commercial statistics package
(Systat/Sygraph for Windows, Evanston, IL,
U.S.A.). Changes of rms and mpf were evaluated
with regression analysis.

Normalization Procedure

The mean mpf and rms of the first 10 s were

used as reference values. All following data were
divided with this reference value, giving an initial
value of 1.0.

RESULTS

The load levels corresponded to about 15, 22.5

and 30% of MVC. All the tested subjects maintained
the load for the desired time.
There was a significant rise of the rms value at
all three load levels, more at the higher load levels
(Figure 2).
The mpf showed no change at the 0 kg load
level, but a significant decline at the two higher
FIG. 1. Test position and hand loads. load levels (Figure 3).

Journal of Electromyography & Kinesiology Vol. 5, No. 4, 1995

 MUSCLE FATIGUE AND EMG MEASUREMENTS 241

2kg hand load

+lS%/min (pcO.OOlJ
1T
1 kg hand load
+lS%/min (p<O.ON)

0 kg hand load
+7%/min (p<O.OOl)

FIG. 2. Fatigue effects on EMG amplitude parameters (rms). Significance test was performed with regression analysis of the data.

2'

2 kg hand load
-Il%/min (p<O .(101)
1 kg hand load
llf -3%/min (pcO.OOJ)

FIG. 3. Fatigue effects on EMG spectral parameters (mpf). Significance test was performed with regression analysis of the data.

DISCUSSION maintained the load throughout the test, and only

Muscular fatigue can have different meanings:
impaired motor performance, increased EMG ampli-
tude for a given muscular load, shift of the EMG
spectrum towards lower frequencies, or impaired
force generation4. In this study all the subjects
EMG aspects of fatigue are considered.
There was a significant influence of fatigue,
indicated in both amplitude and spectral parameters.
At high load levels the rise of rms is not only
statistically significant, but the magnitude is also
practically important. There are considerable differ-

Journal of Electromyography & Kinesiology Vol. 5, No. 4, 1995

242 T. O'BERG

ences between different individuals and also within very much if the rise in amplitude is due to an
the same individual if the measurement is repeated increased external load, or if it is due to fatigue.
on several occasions. To overcome some of these In both cases, it is an expression of increased
variations, data are often normalized, i.e. expressed physiological load.
relative to this calibration value. Such a procedure
may increase the comparability of data, but the CONCLUSIONS
results are critically dependent on the stability of
the calibration value. If there is much variation in The fatigue effects, especially on amplitude para-
the calibration value, it may make the situation even meters derived from the trapezius muscle, are of
worse. Mirkai examined normalization routines for such a magnitude that they may jeopardize both the
the amplitude EMG in trunk flexion movements. calibration measurement and the later estimations of
He found large errors if single EMG values were muscular load. It is therefore important that the
used for normalization. Gberg et a1.15 examined muscle is not fatigued during the measurement -
calibration procedures for spectral parameters and neither at calibration, nor at later measurements of
recommended repeated measurements to increase muscular load levels. To obtain some indication of
the stability of the calibration value. muscular fatigue, it is recommended that amplitude
To my knowledge the influence of fatigue on measurements should not be performed without the
calibration values in amplitude EMG has not been simultaneous calculation of spectral parameters.
reported before. As has been shown in this paper,
there are statistically significant and, at high load
REFERENCES
levels, also practically significant changes of both
amplitude and spectral parameters due to muscular 1. Arvidsson A: A statistical method for detecting disturbances
fatigue. The EMG amplitude can increase either and nonstationary events in stochastic signals with special
reference to data selection in EEG and EMG analysts.
due to an increased muscular load, or due to Technical report no. 126. Thesis, Chalmers University of
muscular fatigue, but it may be impossible to Technology, Gothenburg, 1982.
differentiate between these effects. Consequently, 2. Basmajian JV, DeLuca CJ: Muscles alive, Their functions
revealed by electromyography (5th edn). Williams and
there is a need for some kind of indicator of Wilkins, Baltimore, 1985.
muscular fatigue when a calibration value is estab- 3. DeLuca CJ: Myoelectrical manifestation of localized muscu-
lished. Spectral parameters are considered to reflect lar fatigue in humans. CRC Crit Rev Biomed Eng 11:251-279,
1984.
muscular fatigue more specifically2*3. For this reason 4. Edwards RHT: Human muscle function and fatigue. In:
spectral parameters should be evaluated together Human muscle fatigue: Physiological mechanisms, ed by
Porter R, Whelan J, Pitman, London, 1981.
with the amplitude parameters. If there are signifi- 5. Hagg G, Suurkula J, Liew M: A worksite method for
cant changes in the spectral parameters, this can shoulder muscle fatigue measurements using EMG test
be an indication of muscular fatigue and the contractions and zero crossing technique. Ergonomics
calibration value must probably be discarded, or at 30:1541-1551, 1987.
6. Kadefors R. Arvidsson A. Herberts P. Petersen I: EMG
least considered with much caution. data processing in ergonomics, with emphasis on studies of
It is impossible to know whether the feelings of shoulder muscle load. In: Computer-aided electromyography,
fatigue originate in the trapezius muscle or in other ed by Desmedt JE, Prog Clin Neurophysiol 10:273-286,
1983.
structures, for example some other muscle or 7. Kadefors R, Sandsjo L, Gberg T: A portable device for
connective tissue structures such as ligaments and ambulatory assessment of workload from electromyographic
signals. In: Computer applications in ergonomics, occu-
tendons. This may explain the lack of mpf changes pational safety and health, ed by Mattila M, Karwowski W,
at low load levels. It may also be explained by Elsevier Science Publ, Amsterdam, pp 433-437, 1992.
recruitment mechanisms: the fatigued motor units 8. Knaflitz M, Merletti R, DeLuca CJ, Cisari C: Motor unit
recruitment order in voluntary and electrically elicited
being replaced or completed with fresh non-fatigued contractions. In: Electromyographical kinesiology, ed by
motor units. This may also explain the rise of the Anderson PA, Hobart DJ, Danoff JV, Elsevier, Amsterdam,
EMG amplitude, despite the lack of mpf changes. pp 321-324, 1991.
An important question is whether the rms must 9. Kogi K, Hakamada T: Frequency analysis of the surface
electromyogram in muscle fatigue. J Sci Labour (Tokyo)
be translated to mechanical load, or if it should 38:51%528, 1962.
stand as a physiological measure of its own of 10. Merletti R, LoConte LR, Orizio C: Indices of muscle
fatigue. J Electromyogr Kinesiol 1:20-33, 1991.
muscular load. The rise of rms in fatigue may reflect 11. Mirka G: The quantification of EMG normalization error.
a loss of efficiency in the production of mechanical Ergonomics 34:343-352, 1991.
force. With such an approach it does not matter 12. Gberg T, Sandsjo L, Kadefors R: EMG mean power

loam01 of Electromyography & Kinesiology Vol. 5, No. 4, 1995

 MVSCLE FATIGUE AND EMG MEASUREMENTS

frequency in non-fatigued trapezius muscle. Eur J Appl
chysiol 611362-369, 1990.
13. Oberg T, Sandsjo L, Kadefors R: Variability of the EMG
mean power frequency: A study on the trapezius muscle. J
Electromyogr Kinesiol 1:237-243, 1991.
14. Gberg T, Sandsjo L, Kadefors R: Arm movement and EMG
mean power frequency in the trapezius muscle: A comparison
between surface and intramuscular recording techniques.
fjlectromyogr Clin Neurophysiol32:87-96, 1992.
15. Oberg T, Sandsjo L, Kadefors R: EMG mean power
frequency and the need for a reference value. Clin Biomech
9~253-257, 1994.
16. Suurkiila J, Hlgg G: Relations between shoulder-neck
disorders and EMG zero crossings shift in female assembly
workers using the test contraction method. Ergonomics
30:155>1564, 1987.

Journal of Electromyography & Kinesiology Vol. 5, No. 4, 1995