a comparative study of emg indices in muscle fatigue

Research ArticleA Comparative Study of EMG Indices in Muscle FatigueEvaluation Based on Grey Relational Analysis during All-OutCycling Exercise

LejunWang 1 YutingWang 1 Aidi Ma1 Guoqiang Ma2 Yu Ye1

Ruijie Li1 and Tianfeng Lu 1

1Sport and Health Research Center Physical Education Department Tongji University Shanghai China2Physical Education and Sports Science Institute of Shanghai Shanghai 200030 China

Correspondence should be addressed to Tianfeng Lu sytyltf126com

Received 27 October 2017 Revised 9 January 2018 Accepted 12 March 2018 Published 16 April 2018

Academic Editor Cho-Pei Jiang

Copyright copy 2018 Lejun Wang et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

The increased popularization of cycling has brought an increase in cycling-related injuries which has been suggested to beassociated with muscle fatigue However it still remains unclear on the utility of different EMG indices in muscle fatigue evaluationinduced by cycling exercise In this study ten cyclist volunteers performed a 30-second all-out cycling exercise after a warm-upperiod Surface electromyography (sEMG) from vastus lateralis muscle (VL) and power output and cadence were recorded andEMG RMS MF and MPF based on Fourier Transform MDF and MNF based on wavelet packet transformation and C(n) basedon LempelndashZiv complexity algorithm were calculated Utility of the indices was compared based on the grey rational grade ofsEMG indices and power output and cadence The results suggested that MNF derived from wavelet packet transformation wassignificantly higher than other EMG indices indicating the potential application for fatigue evaluation induced by all-out cyclingexercise

1 Introduction

The increased popularization of cycling in transportationrecreation and competition has brought an increase incycling-related injuries [1] It was found that about 42 to65 of recreational cyclists may experience overuse kneepain [1 2] The incidence of all nontraumatic injuries amongcyclists may reach 85 [3] As muscle fatigue may change thekinematics and muscle activation patterns so as to maintaintarget performance the injuries has been suggested to becaused by biomechanical alterations associated with musclefatigue [4 5] which indicate that muscle fatigue monitoringand assessment may be helpful in protocol arrangement toreduce injuries during cycling exercise

Surface electromyography (sEMG) has been widely usedin muscle fatigue evaluation due to its noninvasiveness realtime and applicability [6] In previous researches manysEMG indices have been suggested and compared in musclefatigue assessment including root mean square (RMS) the

median (MF) and mean power frequencies (MPF) basedon Fourier Transform [6] However these researches havebeen documented especially for isometric and isokineticcontraction conditions and studies have rarely focused on theEMG-based muscle fatigue evaluation in cycling exercise Asa result on the utility of different EMG indices in musclefatigue evaluation induced by cycling exercise still remainsunclear

Muscle fatigue has been defined and quantified by thereducedmaximumcapacity to generate force or power output[7] Particularly the decrease of maximal force or poweroutput has been used as the valid criterion to evaluate theutility of other approaches (eg sEMG) in muscle fatigueassessment [8 9] In previous researches the proximity ofresearched indices changes and the decrease of maximalforce or power output has been compared and adoptedto evaluate the utility of indices [10] while grey relationalanalysis is a method to quantifying the proximity of changingtrends between inspected sequences and standard sequence

HindawiBioMed Research InternationalVolume 2018 Article ID 9341215 8 pageshttpsdoiorg10115520189341215

2 BioMed Research International

by grey relational grade [11 12] Therefore the utility of EMGindices inmuscle fatigue can be evaluated by comparing theirchanging trends with maximum force or power output usinggrey relational analysis

This work aims at comparing the utility of sEMG indicesin assessing muscle fatigue induced by all-out cycling exer-cise EMG indices representing fatigue were calculated andgrey rational grade of EMG indices and power output werecalculated and compared The research was performed onthe vastus lateralis muscle (VL) as it has been suggested tobe the principal power producers during cycling exercise [1314] Results were expected to provide improved EMG-basedmethods in muscle fatigue assessment for all-out cyclingexercise to help reduce injuries of cyclists

2 Materials and Methods

21 Participants Seven male and three female cyclist vol-unteers (age 2150 plusmn 467 years height 17500 plusmn 825 cmand weight 7540 plusmn 1091 kg) participated in this studywhich was approved by the Ethics Committee of TongjiUniversity The subjects were all healthy with no knownneuromuscular disorders or musculoskeletal injuries andhad not participated in strenuous physical activity in 24 hoursbefore experiment

22 Experimental Protocol The experiment was conductedin laboratory with the indoor temperature of about 24∘Cand comprised a warm-up exercise and a test exercise Allthe exercises were performed on an air-braked ergometer(Wattbike ProWattbike Ltd Nottingham United Kingdom)that allows the resistance to be set between 1 and 10 levelsTheWattbike measures the forces applied to the chain over a loadcell and angular velocity of the crank twice per revolution tocalculate power output at a rate of 100Hz Power output andcadence during track cycling were measured using SRM pro-fessional power cranks (Schoberer Rad-Messtechnik JulichGermany) To ensure accurate measures a static calibrationprocedure was conducted before the study for both devices[15]

The warm-up exercise consisted of a 5min cycling exer-cise with the air resistance on the ergometer set at level 3and the cadence at 90 rpm followed by a complete rest of3min before the beginning of the test Based on previoustests with the subjects the air resistance on the Wattbikeergometer was set to the level at which the subject mayproduce the maximum power output during all-out cyclingexercise According to this criterion the air resistance on theergometer was set to level 10 for sevenmale cyclists and level 6for three female cyclists According to the calibration reportlevels 6 and 10 of air resistance on the Wattbike ergometerresult in power outputs of 45 and 55W at a cadence of 40 rpmand 785 and 1045W at a cadence of 130 rpm Subjects wereasked to produce the highest possible power output for 30seconds and were verbally encouraged throughout the trials

23 EMG Measurement Surface electromyographic signalswere recorded with three round bipolar AgAgCl electrodesof ME 6000 P8 Surface EMG acquisition instrument (Mega

Electronics System Finland) Electrodes were placed overthe belly of right vastus lateralis (VL) with center-to-centerelectrode distance setting to 2 cm The skin was shaved andcleaned with alcohol wipes before the electrodes were fixedMedical adhesive tape and plastic casts were applied to fix theelectrodes Raw EMG signals were amplified simultaneouslydigitized and acquired by the MegaWin system (MegaElectronics System Finland) at a sampling rate of 1 kHz

24 EMG Data Processing EMG signals recorded from VLwere band-pass filtered at 5ndash500Hz using a 4th-order zero-phase-shift Butterworth filter and were divided into every3-second epochs For each epoch EMG RMS MF (medianfrequency) and MPF (mean power frequency) based onFourier Transform MDF and MNF based on wavelet packettransformation and C(n) based on LempelndashZiv complexityalgorithm were calculated

RMS is defined as

RMS = radicsum119899119894=1 1003816100381610038161003816rawData1198941003816100381610038161003816119899 (1)

where 119894 represents the order number of the dealing samplepoint rawData119894 is the value of the 119894th sample point and 119899 isthe total number of the data points

MF and MPF are defined as follows

intMF

0

119878 (119891) 119889119891 = intinfinMF119878 (119891) 119889119891 = 12 int

infin

0

119878 (119891) 119889119891MPF = int

infin

0119878 (119891) sdot 119891 sdot 119889 (119891)intinfin0119878 (119891) sdot 119889119891

(2)

where 119891 is the frequency 119878(119891) is the power at frequency 119891and 119889(119891) is the frequency resolution

Wavelet packet transformation was employed to analyzesEMG and a wavelet that was a member of the Daubechiesfamily (order 6) was implemented in this analysis On thisbasis MDF and MNF were calculated MDF and MNF aredefined as

intMDF

0

119875 (119905 120596) 119889120596 = intinfinMDF119875 (119905 120596) 119889120596

= 12 intinfin

0

119875 (119905 120596) 119889120596MNF = int

infin

0120596119875 (119905 120596) 119889120596intinfin0119875 (119905 120596) 119889120596

(3)

where 119875(119905 120596) represents the power spectrum of EMG signalsbased on wavelet packet transformation

LempelndashZiv complexity was calculated based on com-plexity C(n) algorithm devised by Kaspar and Schuster [16]and its value is between 0 and 1

25 Grey Rational Grade Calculation In the grey relationalgrade calculation EMG indices were selected as inspectedsequences while power (or cadence) was chosen as standard

BioMed Research International 3

Table 1 Grey relational grade between EMG indices and pedaling performance

RMS MF MPF MDF MNF C(n)Power 047 plusmn 006 070 plusmn 006 071 plusmn 003 068 plusmn 006 078 plusmn 005 056 plusmn 009Cadence 043 plusmn 009 070 plusmn 006 069 plusmn 006 067 plusmn 008 074 plusmn 005 047 plusmn 007

sequence Each data of inspected sequences and standardsequence was normalized by dividing the average valueof each sequence Then the grey relational coefficient wascalculated using Dengrsquos grey relational grade formula

corr (1199090 (119896) 119909119894 (119896)) = Δmin + 119901ΔmaxΔ 0119894 (119896) + 119901Δmax (4)

where

(1) 119894 = 1 2 3 119898 119896 = 1 2 3 119899(2) 1199090 is standard sequence and 119909119894 is inspected sequence(3) Δ 0119894 = 1199090(119896)minus119909119894(119896) is the difference between 1199090 and119909119894(4) Δmin = forallminmin

119894forall1198961199090(119896) minus 119909119894(119896) Δmax =forallmaxmax

119894forall1198961199090(119896) minus 119909119894(119896)

(5) 119901 is distinguishing coefficient and 119901 isin [0 1] In thisstudy we took 119901 = 05 according to previous research[11]

When the grey relational coefficient is calculated themean value of the grey relational coefficient is taken as thegrey relational grade

CORR (1199090 119909119894) = 1119899119899sum119896=1

corr (1199090 (119896) 119909119894 (119896)) (5)

Grey relational grade of CORR in this study ranged from0 to 1 A larger value of CORR indicates a more proximity ofchanging trends between EMG index and power output andthus a better utility to evaluate muscle fatigue

Data processing was performed using MATLAB R2016asoftware (Mathworks USA)

26 Statistical Analysis The statistical analysis was per-formed using SPSS 130 for windows (SPSS Inc Chicago ILUSA) Normality was tested using the Kolmogorov-Smirnovtest One-way repeated-measures variance analysis was usedto determine the difference of power cadence and EMGindices in different pedaling phases Two-factor varianceanalysis was used to compare the grey relational grade ofdifferent EMG indices and pedaling performance (power andcadence) All significance thresholds were fixed at 120572 = 0053 Results

Examples of power output cadence and raw EMG signalsof vastus lateralis are shown in Figure 1 It can be observedfrom the figure that during the 30-second all-out cyclingexercise subjects reached their maximum value of poweroutput and cadence in approximately the 4th and 6th second

of the exercise respectively Power output and cadence beganto decrease progressively once they reach the peak value inthe later exercise

Figure 2 shows the average power output (a) and cadence(b) of all subjects calculated for every 3-second duringcycling exercise The power reached the maximum valueof 93950 plusmn 21206W in the second 3-second epoch anddeclined to 48150 plusmn 10556W in the last 3-second epochCorrespondingly the cadence reached 14450 plusmn 495RPMin the second 3-second epoch and decreased to 11537 plusmn463RPM in the last 3-second epoch One-way repeated-measures variance analysis results revealed that power outputand cadence had significant difference in 10 pedaling periods(power 119865 = 32858 119875 le 0001 cadence 119865 = 46705119875 le 0001) Power output and cadence showed a progressivelyapproximate linear decrease tendency from the third 3-second cycling exercise

Figure 3 showed the average EMG RMS C(n) MFMPF MDF and MNF of all subjects calculated for every 3-second during cycling exercise The RMS C(n) MF MPFMDF and MNF in the first and last 3-second epoch were2212689 plusmn 1780551 120583V 5580 plusmn 645Hz 11982 plusmn 777Hz10183plusmn523Hz 13749plusmn802Hz 061plusmn006 and 1763563plusmn1168706 120583V 4895plusmn406Hz 10473plusmn737Hz 9272plusmn805Hz12179 plusmn 782Hz and 056 plusmn 008 respectively During thepedaling exercise the EMG RMS C(n) MF MPF MDFand MNF all showed a declining tendency while EMG MFMPF MDF and MNF showed a more significant decreasedtendency compared with EMG RMS and C(n) One-wayrepeated-measures variance analysis results revealed that theduration time factor had significant influence on EMG RMSMF MPF MDF MNF and C(n) (RMS 119865 = 2957 119875 le 005MF 119865 = 6315 119875 le 0001 MPF 119865 = 13114 119875 le 0001MDF 119865 = 3969 119875 le 0001 MNF 119865 = 8915 119875 le 0001C(n) 119865 = 6087 119875 le 0001) Post hoc analysis showed thatmost of MF MPF MDF and MNF values were significantlydecreased compared to the value of precedent

Table 1 shows the grey relational grade between EMGindices and pedaling performanceThe grey relational gradesbetween EMG indices of RMS MF MPF MDF MNF C(n)and power were 047plusmn006 070plusmn006 071plusmn003 068plusmn006078 plusmn 005 and 056 plusmn 009 while the grey relational gradesbetween EMG indices and cadence were 043 plusmn 009 070 plusmn006 069 plusmn 006 067 plusmn 008 074 plusmn 005 and 047 plusmn 007

The statistical analysis revealed significant effects of bothinspected sequences (119865 = 68241 119875 le 0001) and standardsequences (119865 = 7348 119875 le 001) on the grey relational gradevalue No significant interaction effect of inspected sequencesby standard sequences was found (119865 = 1207 119875 gt 005)Post hoc analysis showed that grey relational grade calculatedbetween EMG indices and power was significantly greaterthan between cadence On the other hand grey relational


5 100 20 25 3015Duration time (s)

400

600

800

1000

1200

1400

Pow

er (w

)

(a)

2015 25 305 100Duration time (s)

200

300

400

500

600

700

800

Pow

er (w

)

(b)

5 10 15 20 25 300Duration time (s)

708090

100110120130140150160

Cade

nce (

rpm

)

(c)

5 100 20 25 3015Duration time (s)

80

90

100

110

120

130

140

150

Cade

nce (

rpm

)

(d)

5 10 15 20 25 300Duration time (s)

minus8000minus6000minus4000minus2000

02000400060008000

EMG

ampl

itude

(V

)

(e)


minus5000

500100015002000

EMG

ampl

itude

(V

)


(f)

Figure 1 Power output cadence and raw EMG signals of vastus lateralis for two representative subjects during 30-second all-out cyclingexercise

272412 15 18 21 303 96Duration time (s)

300400500600700800900

100011001200

Pow

er (w

)

(a)

6 9 12 15 18 21 24 27 303Duration time (s)

90

100

110

120

130

140

150

160

Cade

nce (

rpm

)

(b)

Figure 2 Average power output (a) and pedaling rate (b) of all subjects calculated for every 3-second during cycling exercise


3 6 9 12 15 18 21 24 27 300Duration time (s)

16500

21500

26500

31500

36500

41500

RMS

(V

)

(a)

3 6 9 12 15 18 21 24 27 300Duration time (s)

054

056

058

06

062

064

066

068

Lem

pelndash

Ziv

com

plex

ity

(b)

3 6 9 12 15 18 21 24 27 300Duration time (s)

46485052545658606264

MF

(Hz)

(c)

3 6 9 12 15 18 21 24 27 300Duration time (s)

100

105

110

115

120

125

130

MPF

(Hz)

(d)

3 6 9 12 15 18 21 24 27 300Duration time (s)

88

93

98

103

108

MD

F (H

z)

(e)

3 6 9 12 15 18 21 24 27 300Duration time (s)

120

125

130

135

140

145

150

MN

F (H

z)

(f)

Figure 3 Average EMG RMS C(n) MF MPF MDF and MNF of all subjects calculated for every 3-second during cycling exercise

grade of MNF was significantly higher than other EMGindices while grey relational grade of RMS and C(n) wassignificantly lower than other four EMG indices (119875 lt 005)4 Discussion

The main purpose of this study was to determine the utilityof several EMG-based fatigue indices in the context of cyclingexercise In the comparative research of EMG-based musclefatigue evaluation previous studies have mainly focused onisometric and isokinetic contraction conditions and rarelyhave focused on the cycling exercise [6 10 17 18] In thepresent study the utility of the EMG indices inmuscle fatigueevaluation was quantified based on the grey rational grade

between EMG indices and power output Results of this studysuggested that grey relational grade of MNF was significantlyhigher than other EMG indices indicating the potentialapplication for fatigue evaluation induced by all-out cyclingexercise

In the previous research [19] we have compared sensitiv-ity and stability of EMG RMS MPF MNF and C(n) in eval-uation of rectus femoris fatigue induced by 60-second all-outcycling exercise and found that MNF have the highest fatiguesensitivity while RMS had the lowest fatigue sensitivity Thesensitivity of MPF and C(n) had no significant differenceReferring to stability C(n) was the optimum index andMNFthe second followed by MPF and RMS In this study theoptimum utility of MNF and inferior utility of RMS have also


been found which is consistent with the previous researchHowever C(n) was not found to have favourable advantagecompared to MF and MPF

RMS showed inferior utility compared to other indicesindicating inconsistent performance to evaluate musclefatigue The result is in agreement with previous studieswhich have revealed the inconsistent changes of RMS duringcontractions [20] As RMS can be easily influenced by exper-iment conditions such as muscle contraction style workloadendurance time and other factors the use of this index asfatigue indicator should be interpreted with caution [10 21]

MF and MPF have been accepted as the representativeindicators of muscle fatigue and have been widely employedin muscle fatigue evaluation in previous researches [2223] It has been reported that both in isometric and indynamic fatiguing contraction conditions MF and MPFshowed significant decreasing tendency [1 24 25] Howeverother studies have shown concern on the use of MF andMPFas fatigue index in dynamic contraction due to the limitationof Fourier Transform in nonstationary signal analysis [26] Inthis study EMG recorded from VL were nonstationary andnonlinear signals due to the obvious electrode shifts relativeto the muscle fiber and changes in the conductivity propertyof the tissues in the strenuous all-out cycling exercise whichmay influence the sensitivity and stability of MF and MPFas fatigue indices [27ndash29] The results showed that the greyrelational grade of MF and MPF were higher than RMS andC(n) and lower thanMNF indicating that MF andMPF werenot the optimum fatigue indices in all-out cycling exercise

Wavelet transformation has been suggested to be suitablefor the analysis of nonstationarymyoelectric signals recordedin dynamic contractions Karlsson et al [30] have shown thatcontinuous wavelet transform has better accuracy in estimat-ing time-dependent spectral moments than those obtainedby using short-time Fourier Transform the WignerndashVilledistribution and the ChoindashWilliams distribution methods[13] Karlsson et al used wavelet transformation to studymovements at different angular velocities and found thatwavelet transform was very reliable for the analysis of non-stationary biological signals [31] In agreement with previousresearches the grey relational grade ofMNFwas significantlyhigher than other parameters indicating the optimum utilityin muscle fatigue assessment

LempelndashZiv complexity was calculated using regressionfunction method and can reflect new pattern generatingspeed of time series along with its length increase andreflects the randomness of the time sequence [32] Previousresearches have demonstrated that the LempelndashZiv com-plexity method is easy to use and effective in revealing thedynamic characteristics and variations of exercise fatigue[33 34] In our previous research LempelndashZiv complexityshowed the highest stability while the sensitivity was lowerthan MPF and MNF [19] Therefore the poor utility ofLempelndashZiv complexity C(n) revealed in this study mayattribute to the low sensitivity of reflecting muscle fatigueand is not suitable for the fatigue index induced by cyclingexercise

Muscle fatigue represents a complex phenomenon andencompasses a number of changes occurring at both central

and peripheral level [35ndash38] Although central and peripheralmechanisms are highly interactive the contribution of twomechanisms may be quite different in certain conditions Forexample previous researches have demonstrated that musclefatigue induced by very high force level mainly occurredat perpetual level while sustained contraction at low forcesproduces prominent central fatigue [39ndash42] In this studysubjects performed very vigorous dynamic cycling exercisewith high force level and no doubt would lead to remarkableperpetual fatigue as well as central fatigue As utility of EMGindices may be quite different in evaluating central fatigueand peripheral fatigue [39ndash42] the difference utility of EMGvariables in muscle fatigue evaluation would be explainedby the different fatigue mechanism of central and perpetualfactors

In conclusion surface EMG recorded fromVL during all-out cycling exercise was nonlinear and nonstationary signalsthat refrain the application of Fourier Transform whileMNF derived from wavelet packet transformation showedthe maximum grey rational grade with both power outputand cadence indicating the potential application for fatigueevaluation induced by all-out cycling exercise

Conflicts of Interest

The authors declare that there are no conflicts of interestregarding the publication of this paper

Acknowledgments

This work was supported by the Fundamental ResearchFunds for the Central Universities (no 1500219130) Researchand Construction Project of Teaching Reform of TongjiUniversity and the Twelfth Experimental Teaching ReformProject of Tongji University and the Scientific Fitness Guid-ance Project of National Sports General Administration ofChina (2017B026)

References

[1] J B Dingwell J E Joubert F Diefenthaeler and J D TrinityldquoChanges in muscle activity and kinematics of highly trainedcyclists during fatiguerdquo IEEE Transactions on Biomedical Engi-neering vol 55 no 11 pp 2666ndash2674 2008

[2] A L Dannenberg S Needle D Mullady and K B KolodnerldquoPredictors of injury among 1638 riders in a recreational long-distance bicycle tour Cycle Across Marylandrdquo The AmericanJournal of Sports Medicine vol 24 no 6 pp 747ndash753 1996

[3] N J Dettori and D C Norvell ldquoNon-traumatic bicycle injuriesA review of the literaturerdquo Sports Medicine vol 36 no 1 pp7ndash18 2006

[4] M Lehnert M De Ste Croix A Zaatar J Hughes R Varekovaand O Lastovicka ldquoMuscular and neuromuscular control fol-lowing soccer-specific exercise inmale youth Changes in injuryrisk mechanismsrdquo Scandinavian Journal of Medicine amp Sciencein Sports vol 27 no 9 pp 975ndash982 2017

[5] F Delvaux J-F Kaux and J-L Croisier ldquoLower limb muscleinjuries Risk factors and preventive strategiesrdquo Science ampSports vol 32 no 4 pp 179ndash190 2017


[6] M Cifrek V Medved S Tonkovic and S Ostojic ldquoSurfaceEMGbasedmuscle fatigue evaluation in biomechanicsrdquoClinicalBiomechanics vol 24 no 4 pp 327ndash340 2009

[7] N K Voslashllestad ldquoMeasurement of human muscle fatiguerdquoJournal of NeuroscienceMethods vol 74 no 2 pp 219ndash227 1997

[8] Y Ikemoto S Demura S Yamaji and T Yamada ldquoComparisonof force-time parameters and EMG in static explosive grippingby various exertion conditions Muscle fatigue state and sub-maximal exertionrdquoThe Journal of Sports Medicine and PhysicalFitness vol 46 no 3 pp 381ndash387 2006

[9] A Troiano F Naddeo E Sosso G Camarota R Merlettiand L Mesin ldquoAssessment of force and fatigue in isometriccontractions of the upper trapezius muscle by surface EMGsignal and perceived exertion scalerdquo Gait amp Posture vol 28 no2 pp 179ndash186 2008

[10] Yassierli and M A Nussbaum ldquoUtility of traditional and alter-native EMG-based measures of fatigue during low-moderatelevel isometric effortsrdquo Journal of Electromyography amp Kinesi-ology vol 18 no 1 pp 44ndash53 2008

[11] K Wen ldquoThe quantized transformation in Dengrsquos grey rela-tional graderdquo Grey Systems Theory and Application vol 6 no3 pp 375ndash397 2016

[12] G D Xu P Guo X M Li and Y Y Jia ldquoGrey relational analysisand its application based on the angle perspective in time seriesrdquoJournal of Applied Mathematics vol 2014 Article ID 568697 7pages 2014

[13] H Akima R Kinugasa and S Kuno ldquoRecruitment of the thighmuscles during sprint cycling by muscle functional magneticresonance imagingrdquo International Journal of Sports Medicinevol 26 no 4 pp 245ndash252 2005

[14] S Bercier R Halin P Ravier et al ldquoThe vastus lateralisneuromuscular activity during all-out cycling exerciserdquo Journalof Electromyography amp Kinesiology vol 19 no 5 pp 922ndash9302009

[15] A Nimmerichter and C A Williams ldquoComparison of poweroutput during ergometer and track cycling in adolescentcyclistsrdquoThe Journal of Strength and Conditioning Research vol29 no 4 pp 1049ndash1056 2015

[16] F Kaspar and H G Schuster ldquoEasily calculable measure forthe complexity of spatiotemporal patternsrdquo Physical Review AAtomic Molecular and Optical Physics vol 36 no 2 pp 842ndash848 1987

[17] P A Karthick and S Ramakrishnan ldquoMuscle fatigue analysisusing surface EMG signals and time-frequency based medium-to-low band power ratiordquo IEEE Electronics Letters vol 52 no 3pp 185-186 2016

[18] S Okkesim and K Coskun ldquoFeatures for muscle fatigue com-puted from electromyogram and mechanomyogram A newonerdquo Proceedings of the Institution of Mechanical Engineers PartH Journal of Engineering in Medicine vol 230 no 12 pp 1096ndash1105 2016

[19] L Wang Y Huang M Gong and G Ma ldquoSensitivity and sta-bility analysis of sEMG indices in evaluating muscle fatigue ofrectus femoris caused by all-out cycling exerciserdquo in Proceedingsof the 4th International Congress on Image and Signal ProcessingCISP 2011 pp 2779ndash2783 China October 2011

[20] T I Arabadzhiev V G Dimitrov N A Dimitrova and GV Dimitrov ldquoInterpretation of EMG integral or RMS andestimates of lsquoneuromuscular efficiencyrsquo can be misleading infatiguing contractionrdquo Journal of Electromyography amp Kinesiol-ogy vol 20 no 2 pp 223ndash232 2010

[21] W Fu Y Fang Y Gu L Huang L Li and Y Liu ldquoShoe cushion-ing reduces impact andmuscle activation during landings fromunexpected but not self-initiated dropsrdquo Journal of Science andMedicine in Sport vol 20 no 10 pp 915ndash920 2017

[22] J A Mutchler J T Weinhandl M C Hoch and B L VanLunen ldquoReliability and fatigue characteristics of a standing hipisometric endurance protocolrdquo Journal of Electromyography ampKinesiology vol 25 no 4 pp 667ndash674 2015

[23] C M Smith T J Housh N D M Jenkins et al ldquoCombiningregression and mean comparisons to identify the time courseof changes in neuromuscular responses during the process offatiguerdquo Physiological Measurement vol 37 no 11 pp 1993ndash2002 2016

[24] L Wang A Lu S Zhang W Niu F Zheng and M GongldquoFatigue-related electromyographic coherence and phase syn-chronization analysis between antagonistic elbow musclesrdquoExperimental Brain Research vol 233 no 3 pp 971ndash982 2014

[25] M S Tenan R G McMurray B Troy Blackburn M McGrathand K Leppert ldquoThe relationship between blood potassiumblood lactate and electromyography signals related to fatigue ina progressive cycling exercise testrdquo Journal of Electromyographyamp Kinesiology vol 21 no 1 pp 25ndash32 2011

[26] T W Beck T J Housh G O Johnson et al ldquoComparisonof Fourier and wavelet transform procedures for examiningthe mechanomyographic and electromyographic frequencydomain responses during fatiguing isokinetic muscle actions ofthe biceps brachiirdquo Journal of Electromyography amp Kinesiologyvol 15 no 2 pp 190ndash199 2005

[27] A Phinyomark P Phukpattaranont C Limsakul and MPhothisonothai ldquoElectromyography (EMG) signal classifica-tion based on detrended fluctuation analysisrdquo Fluctuation andNoise Letters vol 10 no 3 pp 281ndash301 2011

[28] R K Mishra and R Maiti ldquoNon-linear signal processingtechniques applied on emg signal for muscle fatigue analysisduring dynamic contractionrdquo in Proceedings of the 22nd CIRPDesign Conference CIRP Design 2012 pp 193ndash203 ind March2012

[29] A O Andrade P Kyberd and S J Nasuto ldquoThe applicationof the Hilbert spectrum to the analysis of electromyographicsignalsrdquo Information Sciences vol 178 no 9 pp 2176ndash21932008

[30] S Karlsson J Yu and M Akay ldquoTime-frequency analysis ofmyoelectric signals during dynamic contractions a compara-tive studyrdquo IEEE Transactions on Biomedical Engineering vol47 no 2 pp 228ndash238 2000

[31] J S Karlsson B Gerdle and M Akay ldquoAnalyzing surfacemyoelectric signals recorded during isokinetic contractionsrdquoIEEE Engineering inMedicine and BiologyMagazine vol 20 no6 pp 97ndash105 2001

[32] C Zhang H Wang and R Fu ldquoAutomated detection of driverfatigue based on entropy and complexity measuresrdquo IEEETransactions on Intelligent Transportation Systems vol 15 no1 pp 168ndash177 2014

[33] M Yijian L Xinyuan and W Tingting ldquoThe L-Z complexityof exercise-induced muscle fatigue based on acoustic myog-raphyerdquo Journal of Applied Physics vol 115 no 3 Article ID034701 2014

[34] M Talebinejad A D C Chan and A Miri ldquoA Lempel-Zivcomplexity measure for muscle fatigue estimationrdquo Journal ofElectromyographyampKinesiology vol 21 no 2 pp 236ndash241 2011

[35] S Slobounov ldquoInjuries in athletics Causes and consequencesrdquoInjuries in Athletics Causes and Consequences pp 1ndash544 2008


[36] C R Carriker ldquoComponents of FatiguerdquoThe Journal of Strengthand Conditioning Research vol 31 no 11 pp 3170ndash3176 2017

[37] S ZhangW Fu J Pan LWang R Xia andY Liu ldquoAcute effectsof Kinesio taping onmuscle strength and fatigue in the forearmof tennis playersrdquo Journal of Science and Medicine in Sport vol19 no 6 pp 459ndash464 2016

[38] L Wang A Ma Y Wang S You and A Lu ldquoAntagonistMuscle Prefatigue Increases the Intracortical Communicationbetween Contralateral Motor Cortices during Elbow ExtensionContractionrdquo Journal of Healthcare Engineering vol 2017 pp 1ndash6 2017

[39] J L Smith P G Martin S C Gandevia and J L TaylorldquoSustained contraction at very low forces produces prominentsupraspinal fatigue in human elbow flexor musclesrdquo Journal ofApplied Physiology vol 103 no 2 pp 560ndash568 2007

[40] G Boccia D Dardanello C Zoppirolli et al ldquoCentral andperipheral fatigue in knee and elbow extensor muscles after along-distance cross-country ski racerdquo Scandinavian Journal ofMedicine amp Science in Sports vol 27 no 9 pp 945ndash955 2017

[41] T D Eichelberger and M Bilodeau ldquoCentral fatigue of thefirst dorsal interosseous muscle during low-force and high-force sustained submaximal contractionsrdquo Clinical Physiologyand Functional Imaging vol 27 no 5 pp 298ndash304 2007

[42] T Yoon B S Delap E E Griffith and S K Hunter ldquoMech-anisms of fatigue differ after low- and high-force fatiguingcontractions in men and womenrdquo Muscle amp Nerve vol 36 no4 pp 515ndash524 2007

Stem Cells International

Hindawiwwwhindawicom Volume 2018


MEDIATORSINFLAMMATION

of

EndocrinologyInternational Journal of



Disease Markers


BioMed Research International

OncologyJournal of



Oxidative Medicine and Cellular Longevity


PPAR Research

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Immunology ResearchHindawiwwwhindawicom Volume 2018

Journal of

ObesityJournal of



Computational and Mathematical Methods in Medicine


Behavioural Neurology

OphthalmologyJournal of


Diabetes ResearchJournal of



Research and TreatmentAIDS


Gastroenterology Research and Practice


Parkinsonrsquos Disease

Evidence-Based Complementary andAlternative Medicine

Volume 2018Hindawiwwwhindawicom

Submit your manuscripts atwwwhindawicom


by grey relational grade [11 12] Therefore the utility of EMGindices inmuscle fatigue can be evaluated by comparing theirchanging trends with maximum force or power output usinggrey relational analysis

This work aims at comparing the utility of sEMG indicesin assessing muscle fatigue induced by all-out cycling exer-cise EMG indices representing fatigue were calculated andgrey rational grade of EMG indices and power output werecalculated and compared The research was performed onthe vastus lateralis muscle (VL) as it has been suggested tobe the principal power producers during cycling exercise [1314] Results were expected to provide improved EMG-basedmethods in muscle fatigue assessment for all-out cyclingexercise to help reduce injuries of cyclists

2 Materials and Methods

21 Participants Seven male and three female cyclist vol-unteers (age 2150 plusmn 467 years height 17500 plusmn 825 cmand weight 7540 plusmn 1091 kg) participated in this studywhich was approved by the Ethics Committee of TongjiUniversity The subjects were all healthy with no knownneuromuscular disorders or musculoskeletal injuries andhad not participated in strenuous physical activity in 24 hoursbefore experiment

22 Experimental Protocol The experiment was conductedin laboratory with the indoor temperature of about 24∘Cand comprised a warm-up exercise and a test exercise Allthe exercises were performed on an air-braked ergometer(Wattbike ProWattbike Ltd Nottingham United Kingdom)that allows the resistance to be set between 1 and 10 levelsTheWattbike measures the forces applied to the chain over a loadcell and angular velocity of the crank twice per revolution tocalculate power output at a rate of 100Hz Power output andcadence during track cycling were measured using SRM pro-fessional power cranks (Schoberer Rad-Messtechnik JulichGermany) To ensure accurate measures a static calibrationprocedure was conducted before the study for both devices[15]

The warm-up exercise consisted of a 5min cycling exer-cise with the air resistance on the ergometer set at level 3and the cadence at 90 rpm followed by a complete rest of3min before the beginning of the test Based on previoustests with the subjects the air resistance on the Wattbikeergometer was set to the level at which the subject mayproduce the maximum power output during all-out cyclingexercise According to this criterion the air resistance on theergometer was set to level 10 for sevenmale cyclists and level 6for three female cyclists According to the calibration reportlevels 6 and 10 of air resistance on the Wattbike ergometerresult in power outputs of 45 and 55W at a cadence of 40 rpmand 785 and 1045W at a cadence of 130 rpm Subjects wereasked to produce the highest possible power output for 30seconds and were verbally encouraged throughout the trials

23 EMG Measurement Surface electromyographic signalswere recorded with three round bipolar AgAgCl electrodesof ME 6000 P8 Surface EMG acquisition instrument (Mega

Electronics System Finland) Electrodes were placed overthe belly of right vastus lateralis (VL) with center-to-centerelectrode distance setting to 2 cm The skin was shaved andcleaned with alcohol wipes before the electrodes were fixedMedical adhesive tape and plastic casts were applied to fix theelectrodes Raw EMG signals were amplified simultaneouslydigitized and acquired by the MegaWin system (MegaElectronics System Finland) at a sampling rate of 1 kHz

24 EMG Data Processing EMG signals recorded from VLwere band-pass filtered at 5ndash500Hz using a 4th-order zero-phase-shift Butterworth filter and were divided into every3-second epochs For each epoch EMG RMS MF (medianfrequency) and MPF (mean power frequency) based onFourier Transform MDF and MNF based on wavelet packettransformation and C(n) based on LempelndashZiv complexityalgorithm were calculated

RMS is defined as

RMS = radicsum119899119894=1 1003816100381610038161003816rawData1198941003816100381610038161003816119899 (1)

where 119894 represents the order number of the dealing samplepoint rawData119894 is the value of the 119894th sample point and 119899 isthe total number of the data points

MF and MPF are defined as follows

intMF

0

119878 (119891) 119889119891 = intinfinMF119878 (119891) 119889119891 = 12 int

infin

0

119878 (119891) 119889119891MPF = int

infin

0119878 (119891) sdot 119891 sdot 119889 (119891)intinfin0119878 (119891) sdot 119889119891

(2)

where 119891 is the frequency 119878(119891) is the power at frequency 119891and 119889(119891) is the frequency resolution

Wavelet packet transformation was employed to analyzesEMG and a wavelet that was a member of the Daubechiesfamily (order 6) was implemented in this analysis On thisbasis MDF and MNF were calculated MDF and MNF aredefined as

intMDF

0

119875 (119905 120596) 119889120596 = intinfinMDF119875 (119905 120596) 119889120596

= 12 intinfin

0

119875 (119905 120596) 119889120596MNF = int

infin

0120596119875 (119905 120596) 119889120596intinfin0119875 (119905 120596) 119889120596

(3)

where 119875(119905 120596) represents the power spectrum of EMG signalsbased on wavelet packet transformation

LempelndashZiv complexity was calculated based on com-plexity C(n) algorithm devised by Kaspar and Schuster [16]and its value is between 0 and 1

25 Grey Rational Grade Calculation In the grey relationalgrade calculation EMG indices were selected as inspectedsequences while power (or cadence) was chosen as standard






where



119894forall1198961199090(119896) minus 119909119894(119896)



CORR (1199090 119909119894) = 1119899119899sum119896=1

corr (1199090 (119896) 119909119894 (119896)) (5)











5 100 20 25 3015Duration time (s)

400

600

800

1000

1200

1400

Pow

er (w

)

(a)


200

300

400

500

600

700

800

Pow

er (w

)

(b)

5 10 15 20 25 300Duration time (s)

708090

100110120130140150160

Cade

nce (

rpm

)

(c)

5 100 20 25 3015Duration time (s)

80

90

100

110

120

130

140

150

Cade

nce (

rpm

)

(d)

5 10 15 20 25 300Duration time (s)


02000400060008000

EMG

ampl

itude

(V

)

(e)


minus5000

500100015002000

EMG

ampl

itude

(V

)


(f)


272412 15 18 21 303 96Duration time (s)

300400500600700800900

100011001200

Pow

er (w

)

(a)

6 9 12 15 18 21 24 27 303Duration time (s)

90

100

110

120

130

140

150

160

Cade

nce (

rpm

)

(b)



3 6 9 12 15 18 21 24 27 300Duration time (s)

16500

21500

26500

31500

36500

41500

RMS

(V

)

(a)

3 6 9 12 15 18 21 24 27 300Duration time (s)

054

056

058

06

062

064

066

068

Lem

pelndash

Ziv

com

plex

ity

(b)

3 6 9 12 15 18 21 24 27 300Duration time (s)

46485052545658606264

MF

(Hz)

(c)

3 6 9 12 15 18 21 24 27 300Duration time (s)

100

105

110

115

120

125

130

MPF

(Hz)

(d)

3 6 9 12 15 18 21 24 27 300Duration time (s)

88

93

98

103

108

MD

F (H

z)

(e)

3 6 9 12 15 18 21 24 27 300Duration time (s)

120

125

130

135

140

145

150

MN

F (H

z)

(f)

















Acknowledgments


References

















































of




Disease Markers



OncologyJournal of





PPAR Research



Volume 2018


Journal of

ObesityJournal of
























where



119894forall1198961199090(119896) minus 119909119894(119896)



CORR (1199090 119909119894) = 1119899119899sum119896=1

corr (1199090 (119896) 119909119894 (119896)) (5)











5 100 20 25 3015Duration time (s)

400

600

800

1000

1200

1400

Pow

er (w

)

(a)


200

300

400

500

600

700

800

Pow

er (w

)

(b)

5 10 15 20 25 300Duration time (s)

708090

100110120130140150160

Cade

nce (

rpm

)

(c)

5 100 20 25 3015Duration time (s)

80

90

100

110

120

130

140

150

Cade

nce (

rpm

)

(d)

5 10 15 20 25 300Duration time (s)


02000400060008000

EMG

ampl

itude

(V

)

(e)


minus5000

500100015002000

EMG

ampl

itude

(V

)


(f)


272412 15 18 21 303 96Duration time (s)

300400500600700800900

100011001200

Pow

er (w

)

(a)

6 9 12 15 18 21 24 27 303Duration time (s)

90

100

110

120

130

140

150

160

Cade

nce (

rpm

)

(b)



3 6 9 12 15 18 21 24 27 300Duration time (s)

16500

21500

26500

31500

36500

41500

RMS

(V

)

(a)

3 6 9 12 15 18 21 24 27 300Duration time (s)

054

056

058

06

062

064

066

068

Lem

pelndash

Ziv

com

plex

ity

(b)

3 6 9 12 15 18 21 24 27 300Duration time (s)

46485052545658606264

MF

(Hz)

(c)

3 6 9 12 15 18 21 24 27 300Duration time (s)

100

105

110

115

120

125

130

MPF

(Hz)

(d)

3 6 9 12 15 18 21 24 27 300Duration time (s)

88

93

98

103

108

MD

F (H

z)

(e)

3 6 9 12 15 18 21 24 27 300Duration time (s)

120

125

130

135

140

145

150

MN

F (H

z)

(f)

















Acknowledgments


References

















































of




Disease Markers



OncologyJournal of





PPAR Research



Volume 2018


Journal of

ObesityJournal of




















5 100 20 25 3015Duration time (s)

400

600

800

1000

1200

1400

Pow

er (w

)

(a)


200

300

400

500

600

700

800

Pow

er (w

)

(b)

5 10 15 20 25 300Duration time (s)

708090

100110120130140150160

Cade

nce (

rpm

)

(c)

5 100 20 25 3015Duration time (s)

80

90

100

110

120

130

140

150

Cade

nce (

rpm

)

(d)

5 10 15 20 25 300Duration time (s)


02000400060008000

EMG

ampl

itude

(V

)

(e)


minus5000

500100015002000

EMG

ampl

itude

(V

)


(f)


272412 15 18 21 303 96Duration time (s)

300400500600700800900

100011001200

Pow

er (w

)

(a)

6 9 12 15 18 21 24 27 303Duration time (s)

90

100

110

120

130

140

150

160

Cade

nce (

rpm

)

(b)



3 6 9 12 15 18 21 24 27 300Duration time (s)

16500

21500

26500

31500

36500

41500

RMS

(V

)

(a)

3 6 9 12 15 18 21 24 27 300Duration time (s)

054

056

058

06

062

064

066

068

Lem

pelndash

Ziv

com

plex

ity

(b)

3 6 9 12 15 18 21 24 27 300Duration time (s)

46485052545658606264

MF

(Hz)

(c)

3 6 9 12 15 18 21 24 27 300Duration time (s)

100

105

110

115

120

125

130

MPF

(Hz)

(d)

3 6 9 12 15 18 21 24 27 300Duration time (s)

88

93

98

103

108

MD

F (H

z)

(e)

3 6 9 12 15 18 21 24 27 300Duration time (s)

120

125

130

135

140

145

150

MN

F (H

z)

(f)

















Acknowledgments


References

















































of




Disease Markers



OncologyJournal of





PPAR Research



Volume 2018


Journal of

ObesityJournal of




















3 6 9 12 15 18 21 24 27 300Duration time (s)

16500

21500

26500

31500

36500

41500

RMS

(V

)

(a)

3 6 9 12 15 18 21 24 27 300Duration time (s)

054

056

058

06

062

064

066

068

Lem

pelndash

Ziv

com

plex

ity

(b)

3 6 9 12 15 18 21 24 27 300Duration time (s)

46485052545658606264

MF

(Hz)

(c)

3 6 9 12 15 18 21 24 27 300Duration time (s)

100

105

110

115

120

125

130

MPF

(Hz)

(d)

3 6 9 12 15 18 21 24 27 300Duration time (s)

88

93

98

103

108

MD

F (H

z)

(e)

3 6 9 12 15 18 21 24 27 300Duration time (s)

120

125

130

135

140

145

150

MN

F (H

z)

(f)

















Acknowledgments


References

















































of




Disease Markers



OncologyJournal of





PPAR Research



Volume 2018


Journal of

ObesityJournal of






























Acknowledgments


References

















































of




Disease Markers



OncologyJournal of





PPAR Research



Volume 2018


Journal of

ObesityJournal of






























































of




Disease Markers



OncologyJournal of





PPAR Research



Volume 2018


Journal of

ObesityJournal of



















a comparative study of emg indices in muscle fatigue

Documents