Changes in EMG During Short Duration Supra

Maximal and Long Duration Sub-maximal Exercise: A

Comparative Study

Vijay Pal Singh

, Dinesh Kant Kumar

, Barbara Polus

, Steve Fraser

and

Sonia Lo Guidice

School of Electrical and Computer Engineering ,

RMIT University, Melbourne 3000, Australia

Health Sciences,

RMIT University, Bundoora 3053, Australia

Medical Sciences,

RMIT University, Bundoora 3053, Australia

Abstract. Surface electromyogram (sEMG) is a non-invasive recording of the

underlying muscle activities. It is used as a measure of the force of contraction,

and changes in sEMG are used as an indicator for localized muscle fatigue. This

paper reports a study undertaken to determine the difference in the change of

sEMG due to fatigue resulting from short time sprint cycling, and long duration

cycling.

This paper reports the results of the experimental study of the two kinds of ex-

ercises i.e. short duration (supra-maximal) and the long duration (sub-maximal).

The results indicate that measure of the spectrum shift was effective in the de-

tection of fatigue in supra-maximal dynamic contraction but was not useful for

fatigue caused due to long duration sub-maximal cycling.

1 Introduction

The development of muscle fatigue during exercise is associated with a decrement in

performance. Mechanisms of muscle fatigue depend on the exercise conditions (eg.

duration and intensity) and the subject’s level of physical ﬁtness. The decrements in

skeletal muscle power output is also related to neural drive reductions that may also lead

to muscle fatigue in prolonged exercise [1]. This leads to the use of electromyogram - a

measure of the electrical activity of the muscle - as being an indicator of muscle fatigue.

Some researchers have attempted to use the electrical activity of muscle and mus-

cle activation using the surface electromyogram (SEMG) to study fatigue [2], [3], and

[4]. SEMG is a result of summation of number of a separate motor unit action poten-

tials in muscles and is dependent on numerous factors such as the rate of stimulation

of the muscle, size of motor units recruited, morphology of the motor units, electrical

properties of the tissues and the presence of any synchronization of the activity of dif-

ferent motor units. The rate of stimulation of the muscle and size of active motor units

is dependent on the force of contraction required to be produced by the muscle. It is a

complex and non-stationary signal with large inter and intra subject variations.

Pal Singh V., Kant Kumar D., Polus B., Fraser S. and Lo Guidice S. (2006).

Changes in EMG During Short Duration Supra Maximal and Long Duration Sub-maximal Exercise: A Comparative Study.

In Proceedings of the 2nd International Workshop on Biosignal Processing and Classiﬁcation, pages 68-75

DOI: 10.5220/0001224000680075

 SciTePress

Research analysis to date aimed at extracting from the SEMG an indication of local-

ized muscle fatigue has been based on the observed shift of the power spectral density

of the SEMG [5], [6]. Several other parametric measures of the SEMG signal have

also been used as a relative indicator of the muscle fatigue phenomenon for an indi-

vidual subject. These include the Root Mean Square (RMS), spectrum analysis (in-

stantaneous, mean and median frequency) and zero crossing rates. When the muscle is

fatigued, a strengthening of low-frequency components and a reduction in intensity of

high-frequency components modiﬁes the spectrum of the SEMG signal, measurable by

parameters such as median frequency. Measure such as the wavelet coefﬁcients provide

time-frequency information and can improve the reliability of classiﬁcation of sEMG

[7] [8].

During cycling, there is a high degree of variation in the magnitude of sEMG in

each cycle. For the purpose of comparing sEMG parameters from any two cycles, the

authors have earlier demonstrated the use of short-term window at the peak activity for

any cycle [9]. The inter-experiment variability of the magnitude and spectrum of sEMG

is extremely large. To overcome this, the authors have proposed the normalization of

the parameters by considering the ratio of the recordings taken near the fatigued and

healthy state of the muscle [9]. This paper reports the comparison of the shift in the

spectrum due to fatigue for long-duration and sprint cycling using the above-mentioned

technique.

2 Methodology

Experiments were conducted where sEMG was recorded while the subjects did their cy-

cling. The experiments involved sprint cycling and long duration cycling. The recorded

signals were analysed to determine the spectral shift due to the onset of fatigue. The

experiments are detailed below:

2.1 Subjects and Exercise

Short duration. Eleven healthy male subjects performed this exercise, but due to data

corruption, SEMG of only seven subjects was analyzed. This exercise comprised of a

short duration (30 second) cycling on Lode - ergometer with customized software to

fatigue the subjects. Subjects were termed fatigued when the power output dropped by

more than 33%.

Long duration. Nine healthy male subjects performed this exercise, but due to data

corruption, SEMG of only seven subjects was analysed. Subjects did cycling exercise

for as long as they could at sub - maximal level on Lode - ergometer with customised

software to fatigue the subjects. Subjects were termed fatigued at the end of the exercise.

Since the duration of each participant was different, hence his or her time duration were

normalized for the analysis purpose.

2.2 Placement of Electrodes

Electrodes were placed at quadriceps. Electrode placement is as shown in the table

1. The skin was lightly abraded using disposable skin defoliator and cleaned with a

swab soaked in alcohol to reduce skin impedance to less than 60 K. Heart-rate was

monitored (Polar, Finland) to ensure safety of the participant. SEMG was recorded

from the three channels (Table 1) using Delysis (USA) SEMG recording system with

ﬁxed inter-electrode distance, and proprietary electrodes.

Table 1. Channel Assignment for different muscles.

Channel 1 Vastus Lateralis (outside thigh muscle - front)

Channel 2 Vastus Lateralis (inside thigh muscle - front)

Channel 3 Rectus femoris (middle thigh muscle - front)

Analysis of the raw SEMG. After identifying envelope of the raw EMG using mov-

ing RMS, and the peak of each cycle, 100 milliseconds of SEMG immediately after the

peak was selected for analysis (ﬁgure 1). The ﬁrst and the last cycle of the recordings

were discarded because of sudden changes taking place during these segments. RMS

and the median frequency were computed from this 100-millisecond window of three

cycles to ﬁnd out the changes occurring in the EMG as a result of fatigue after the sprint

and long duration exercise. The window of 100-millisecond was chosen because EMG

can be assumed to be stationary for this small amount of time period.

An average was computed for each of the two conditions for each participant. Using

this, ratio of the pre and post RMS and MF was computed for each subject and for the

three channels. A ratio less than one would indicate a decrease due to fatigue. Pair-

wise t - test using online software by SISA was conducted to evaluate the statistical

signiﬁcance of the results, i.e. change in the parameters in before and after fatigue

scenario.

3 Results and Observations

3.1 RMS

Figure 2 shows the ratio of the RMS of three channels in short duration exercise as

depicted by table 1. Figure 3 shows the same ratio but for the long duration exercise.

Ratios are taken by dividing the value of RMS near the end of the exercise (where

subjects are fatigued) by the value of RMS at the start of the exercise (where subjects

are not fatigued). From these ﬁgures and the t test results in the statistical analysis

section, we can observe that there is no signiﬁcant change in the RMS in the both

Fig. 1. Windowed Raw Signal (Illustration only) from Vastus Lateralis of one of the subjects.

cases. For short duration it can be attributed to the fact that the dynamic contraction

was supra maximal right from the start and recruitment of additional motor units was

not possible since increase in RMS is attributed to the increase in number of recruited

motor units. There was no signiﬁcant change in RMS in long duration exercise as well.

Possible reason for this may be that the load on the subject was not increased during the

exercise. The recruitment strategy by the peripheral nervous system may be such that

in the wake of unchanged load it may be switching between different MUs to maintain

a constant force until muscles ran out of energy stores.

3.2 Median Frequency

In a similar way as RMS, median frequency was calculated for the 100-millisecond

window from the peak activity of each cycle. In the end the ratio of the post and pre

fatigue value was taken to ﬁnd out the change occurring in the median frequency. Figure

4 below shows the ratio of the median frequency in short duration while the ﬁgure 5

shows the same for the long duration exercise. Together with ﬁgure 4 and table 2 it can

be very well seen that there is a signiﬁcant decrease in the median frequency during

the short duration supra-maximal exercise. This can be attributed to the physiological

changes that happen relatively at faster rate due to subject being fatigued in only 30

seconds. As a result of this there is strengthening of lower frequency components of the

surface EMG. Contrary to this there is no signiﬁcant change in the median frequency in

longer duration exercise as it is clear from the ﬁgure 5 and table 3.

Fig. 2. Ratio of the RMS of the 100 ms window slice from the three muscles VL (1), VM (2) and

RF (3) in short duration exercise.

Fig. 3. Ratio of the RMS of the 100 ms window slice from the three muscles VL (1), VM (2) and

RF (3) in long duration exercise.

Fig. 4. Ratio of Median Frequency of the 100 ms window slice from the three muscles VL (1),

VM (2) and RF (3) in the short duration exercise.

Fig. 5. Ratio of Median Frequency of the 100 ms window slice from the three muscles VL (1),

VM (2) and RF (3) in the long duration exercise.

3.3 Statistical Analysis

All results obtained were tested statistically to ﬁnd out the signiﬁcance of change in

the RMS and median frequency following fatigue. Online software by SISA was used

to perform pair-wise t test on the data. T test results done on the RMS and median

frequency for the short duration are shown in the table 2 and for the long duration they

are indicated in the table 3 below.

Table 2. T Test for median frequency and RMS in the short duration exercise.

Median Frequency RMS

Channel T P Signiﬁcance T P Signiﬁcance

1 3.07 0.01 95% 0.37 0.36 28%

2 3.19 0.01 98% 0.67 0.73 74%

3 2.05 0.14 91% 0.49 0.68 69%

Table 3. T Test for median frequency and RMS in the long duration exercise.

Median Frequency RMS

Channel T P Signiﬁcance T P Signiﬁcance

1 1.16 0.14 86% 0.39 0.35 64%

2 0.88 0.21 79% 0.74 0.24 76%

3 1.28 0.12 87% 0.33 0.38 62%

4 Discussions

From the results above, it is evident that the changes in the muscle activity as measured

using sEMG are entirely different for the long duration and sprint of exercise. RMS of

sEMG remained unchanged in both the cases. In the short duration exercise the possible

explanation can be that all the motor units were recruited right from the start. The lack

of extra motor units to be recruited when needed results in the unchanged RMS. In long

duration it can be explained based on the contraction of muscles being sub-maximal

and load remains unchanged for the duration of the exercise. This would indicate that

the size and activation of motor units required remains unchanged.

The drop in the median frequency was signiﬁcant in the short duration exercise

while not signiﬁcant for the long duration. The drop in the median frequency for the

sEMG is attributed to the synchronisation of motor unit activity and reduction in con-

duction velocity. The results suggest that during long duration exercise, this may not be

the phenomenon and may be explained on the basis that there is sufﬁcient time for the

neuromotor system to adapt.

5 Conclusion

The study demonstrates that there is a shift in the spectrum of sEMG due to the onset

of muscle fatigue caused due to sprint cycling. Using short-time window at the peak of

the muscle activity, it is possible to identify this shift.

The results also indicate that this shift is not evident when the fatigue is a result of

long-duration cycling. This indicates that the underlying phenomena due to long dura-

tion and sprint cycling appear to be very different. Detailed work that would identify

the biochemical changes due to the two causes of fatigue are required to be undertaken.

References

1. G Kamen, G Caldwell: Physiology and Interpretation of the Electromyogram. Journal of Neu-

rophysiology. 13 366-384, (1996)

2. A V Nargol, A P Jones, P J Kelly, C Greenough: Factors in the reproducibility of electromyo-

graphic power spectrum analysis of lumbar Para spinal muscle fatigue Spine. 24 (9)(1999)

883-888

3. C Oliver, K Tillostson, A P Jones, R Royal : Reproducibility of lumbar Para spinal surface

electromyogram power spectra. Clin. Biomech 11(6) (1996) 317-321

4. N Pah, D Kumar and P Burton: Adding Wavelet Decomposition to Neural Networks for the

Classiﬁcation of Fatigue SEMG. 2nd Conference of the Victorian chapter of the IEEE engi-

neering in medicine and biology society, February 2001

5. C J De Luca: Myoelectric manifestation of localized muscle fatigue in humans. CRC critical

review in Biomedical Engineering. 11 (4) 251-279

6. J R Cram, G S Kasman, and J Holtz: Introduction to Surface Electromyography, Aspen Pub-

lishers, Gaithersburg, Maryland(1998)

7. D Kumar, N Pah, and A Bradley: Wavelet Analysis of Surface Electromyography to Determine

Muscle Fatigue. IEEE transactions on neural systems and rehabilitation engineering.11(4) 400

- 406 (2003)

8. D Kumar, N Pah: Neural Networks and Wavelets for EMG analysis. J EMG and Clinical

Neuro.40 (7) 411- 421(2001)

9. Singh, V.P. Kumar, D.K, et. al.: Strategies to identify muscle fatigue from SEMG during cy-

cling. Intelligent Sensors, Sensor Networks and Information Processing Conference, Proceed-

ings of the 2004, (2004 )547-551