luttmann 2000 (artigo jasa)

International Journal of Industrial Ergonomics 25 (2000) 645}660

Electromyographical indication of muscular fatiguein occupational "eld studies

Alwin Luttmann*, Matthias JaK ger, Wolfgang Laurig

Institute for Occupational Physiology, University of Dortmund, Ardeystra}e 67, D-44139 Dortmund, Germany

Received 1 April 1999; received in revised form 10 August 1999; accepted 29 August 1999

Abstract

Surface electromyography o!ers a valuable tool for the indication of muscular fatigue in occupational "eld studies. Forthis purpose, the time course of the electromyogram (EMG) has to be analysed, in order to detect typical fatigue-inducedchanges such as an increase in EMG amplitude or a shift in the spectral distribution towards lower frequencies. Suchprocedures need a detailed knowledge about the actual activity of the person and muscle under test for the total workingtime. This can be attained by encoding the activity of the person during the work and recording an electrical code signaltogether with the electrophysiological signals. For the indication of muscle fatigue, EMGs for situations connected withsimilar muscle load should only be compared, since the EMG amplitude as well as the spectrum do not only depend uponthe fatigue state, but also upon the produced muscle force. This demand can be ful"lled by (i) interrupting the work andperforming test contractions of known force in a prede"ned body posture or (ii) by comparing situations connected witha certain reference activity. (iii) In a recently developed approach for the joint analysis of EMG amplitude and spectrum(JASA) changes in the amplitude and the spectrum are considered simultaneously. This method permits the discrimina-tion between fatigue-induced and force-related EMG changes. Using this procedure, changes in the EMG can beattributed to categories like fatigue or recovery as well as increase or decrease in the force production of the muscle undertest. Applications from "eld studies during manual materials handling in a weaving mill, price recording at scannercheckouts in a supermarket and the performance of surgical work using endoscopic operation techniques in urologydemonstrate the appropriateness of electromyography for fatigue indication in occupational physiology and ergonomics.Nevertheless, the commonly used measures of muscular fatigue such as increase in EMG amplitude and left shift in EMGspectrum are primarily related to the electrical activation and its propagation along the muscle "bres. Their connectionto the fatigue-induced reduction in the force generating capacity of the muscle under test includes complex physiologicalimplications. Therefore, the need for further development of fatigue indicators which are more directly related tomuscular force is recognized.

Relevance to industry

In occupational health and ergonomics indication of muscular fatigue is needed, since activities inducing muscularfatigue can be performed for a limited time, only, and the quality of work can be in#uenced negatively. Electromyographyo!ers valuable tools for the indication of fatigue and the appropriate assessment of ergonomic design measures. ( 2000Elsevier Science B.V. All rights reserved.

Keywords: Muscular fatigue; Electromyography; EMG amplitude; EMG spectrum; Fatigue indication; Occupational "eld studies

*Corresponding author. Tel.: #49-231-1084-376; fax: #49-231-1084-402.E-mail address: [email protected] (A. Luttmann)

0169-8141/00/$ - see front matter ( 2000 Elsevier Science B.V. All rights reserved.PII: S 0 1 6 9 - 8 1 4 1 ( 9 9 ) 0 0 0 5 3 - 0

1. Introduction

During sustained or repetitive muscle contrac-tions typical changes in surface electromyograms(EMG), such as an increase in the amplitude ora shift in the frequency spectrum towards lowerfrequencies, can be observed. In occupationalelectromyography such EMG changes are com-monly interpreted as signs of muscular fatigue andused to establish the occurance of fatigue. Since,however, the EMG amplitude as well as its spectralcontent do not only depend on the fatigue state,but also on the force production of the muscleunder test, changes in the amplitude and spectrumcannot be unequivocally attributed to muscularfatigue.

Under laboratory conditions it is normally pos-sible to control the force produced by the muscleunder test. Under such circumstances the force canbe kept constant at a known level and a changein the respective EMG can be ascribed to a changein the fatigue state of the muscle. Under "eld condi-tions, however, the force production is determinedby the actual necessities of the work and cannot becontrolled by the investigator. In a general case, it istherefore not possible to decide whether a temporalvariation of an EMG is caused by a change in theforce production or in the fatigue state. Neverthe-less, the indication of muscular fatigue is possibleunder shop-#oor conditions if special requirementsare satis"ed.

One possible way is to compare the EMGs forsituations in which the muscle force is identical.Two methods which ful"l this demand were pre-viously applied in occupational studies: (i) theexecution of test contractions of known force ina de"ned posture, and (ii) the comparison of EMGsfor reference activities of comparable workload. Inanother approach (iii) amplitude and spectral shiftsare considered simultaneously in order to discrimi-nate between force-related and fatigue-inducedEMG changes. To this end a method for the `jointanalysis of EMG spectrum and amplitude (JASA)awas developed and applied (Luttmann et al.,1996a,b; Luttmann and JaK ger, 1998). Using thisprocedure time-related EMG changes can be as-signed to di!erent categories such as fatigue, recov-ery, force increase, or force decrease.

The aim of this paper is to present severalmethods for the indication of muscular fatigue andto weigh their appropriateness using examples fromvarious shop-#oor studies.

2. Methodology for the identi5cation of muscularfatigue in occupational 5eld studies

2.1. Preliminary remarks

In the context of ergonomics and work-physi-ology muscle fatigue is understood as a reductionin the force generating capacity of a muscle whichoccurs in the course of the work. A direct fatiguedetermination based on the mechanical perfor-mance capacity of a muscle needs the repeatedexecution of maximum voluntary contractions atseveral moments of time during the work. Sucha procedure should not be applied in real worksituations, since the measuring strongly disturbsthe work #ow. Furthermore, the measuring is retro-active in its nature, because the execution of testcontractions reduces the performance capacity.Therefore, in occupational studies an indirect fa-tigue determination using electromyography is pre-ferred. All electromyographical fatigue analyses arebased on the assumption that the change in themechanical performance capacity is re#ected incorresponding changes in the myoelectrical signalof the muscle under test. Many studies performedsince the beginning of this century (e.g. Piper,1909,1912; Cobb and Forbes, 1923) have shownthat fatiguing contractions of a muscle coincidewith typical changes in the EMG time course, likean increase in the EMG amplitude or a shift in theEMG spectrum towards lower frequencies. In or-der to detect those changes, long-term EMG re-cordings of selected muscles or muscle groups areperformed for certain periods of time up to totalworking shifts or considerable shift sections. Indata evaluation it has to be examined whethertypical fatigue-induced changes in the EMG can befound.

The preferable method for the measurement ofEMGs in occupational settings is surface electro-myography, since this method allows a noninvasiverecording of myoelectrical signals and a minimal

646 A. Luttmann et al. / International Journal of Industrial Ergonomics 25 (2000) 645}660

Fig. 1. Schematic description of the equipment used to record electrophysiological signals and an action code in ergonomic "eld studies.

restraint of the subjects. The EMGs can be storedusing portable recorders or the signal can be trans-mitted using a portable telemetric device. Espe-cially for the use of portable recorders, datareduction is sometimes advisable. For this purposecharacteristics representing the averaged amplitudelike Root Mean Square (RMS) and Electrical Ac-tivity (EA) or spectral parameters like Mean PowerFrequency (MPF) and Median Frequency (MF)are calculated from the raw EMGs. In "eld appli-cations the recording of raw EMGs should bepreferred, since recordings subjected to prior datareduction have the disadvantage that disturbancescaused, for example, by strong electrode move-ments or power-line hum cannot be distinguishedfrom the electrical signal originating at the muscle"bres.

2.2. Test contractions

It was mentioned before, that an increase in theEMG amplitude may be caused by an increase inthe force production of the muscle under test or bymuscular fatigue. Therefore, an indication of fa-tigue based on the amplitude change is only pos-sible, if situations with identical force production ofthe muscle are compared. This demand can beful"lled using test contractions of prede"ned forcelevels. An appropriate method was developed andapplied in "eld studies by HaK gg et al. (1987). There-by the occupational activity was interrupted atprede"ned moments of time and muscle contrac-tions of known force were performed in a certainposture. EMGs were measured during these test

contractions and the corresponding amplitudeswere compared. Possible EMG changes over timeobserved under such conditions were attributed tothe occurence of muscle fatigue. The use of thismethod needs an exact control of the body postureduring the performance of the test contractions.

2.3. Selection of reference activities of comparableworkload

In an analogous approach EMGs can be com-pared for selected periods of time which are charac-terized by a de"ned activity like holding or liftinga certain object, taking a de"ned body posture,or performing a special movement. The respectiveactivities represent regular elements of the oc-cupational work. They are taken as `referenceactivitiesa comprising a certain muscular forceproduction. By contrast to the method describedabove, customary activities are used instead of testcontractions and the interruption of work is avoid-ed. When applying the reference task method activ-ities should be preferred which are performedrepeatedly during the execution of the work.

To meet this methodological approach a carefuldocumentation of the sequence of operations isneeded which enables, in the data evaluation pro-cess, a precise assignment of the subject's activity tothe EMG. For this purpose a method for the pro-duction of an `electronic protocola was developedand proven in practice (BalleH et al., 1982;Luttmann et al., 1988). A schematic representationof the concept and the equipment is shown in Fig. 1.The subject is accompanied during the total

A. Luttmann et al. / International Journal of Industrial Ergonomics 25 (2000) 645}660 647

working period by an observer who uses a numer-ical keyboard to encode the actual activity accord-ing to a prede"ned code list. The code list must beadopted appropriately for each "eld application.To this end actual characteristics of the work-loadlike body posture of the subject or mass and form ofa manipulated object have to be considered. Theelectrical output signal of the keyboard representsan `action codea which is recorded together withthe electromyograms and } if needed } otherphysiological signals like an electrocardiogram onthe same recording device (computer memory me-dia, magnetic tape). The coding system permits therecording of informations regarding the actual loadat a given moment and to combine those load-related data with the current EMGs. In case ofnon-stationary workplaces it is favorable to usewireless transmitters to relay the action code aswell as the physiological signals to the recordingunit.

During o!-line data evaluation, the electrical ac-tion code enables combining activity data withelectromyographical "ndings and determiningthe myoelectrical activity during particular activ-ities. The code signal can be applied in computer-assisted data evaluation in order to identify allperiods during the shift connected with the sameactivity. With respect to the electromyographicalfatigue analysis this method o!ers a valuable pro-cedure to select sections from EMG time functionswhich are associated with similar workload andpresumably equal muscular force production. Forthe fatigue indication, EMGs of all selected sectionsare compared, and a possible increase in the ampli-tude or a shift in the frequency spectrum towardslower values over time can be attributed to theoccurance of muscular fatigue.

2.4. Joint analysis of EMG spectrum and amplitude(JASA)

In the aforementioned procedures for fatigueindication, EMG changes in the amplitude orspectral domain are considered separately andindependently. Due to the twofold dependency ofthe EMG amplitude and spectrum on force andfatigue, the methods for fatigue indication are onlyapplicable if EMG sections with identical force

production are taken into account. This results ina considerable restriction for use in occupational"eld studies. In the newly developed method for`joint analysis of EMG spectrum and amplitude(JASA)a this restriction is reduced. For this purposethe methodology for fatigue indication was en-hanced and both, the change in the amplitude andthe spectrum were considered simultaneously (Lutt-mann et al., 1996a).

The JASA method is based on the principallyknown relationships between muscular force pro-duction and fatigue state, on the one hand, and theEMG amplitude and spectrum, on the other hand.These relationships were investigated in numerousstudies; with regard to the EMG amplitude it isexperimentally veri"ed that the amplitude increaseswith increasing force as well as with the occurenceof fatigue. A uniform functional relationship be-tween the mechanical force production and theEMG amplitude is not entirely settled (Kumar1996). The spectrum also depends on fatigue andforce production: In case of fatigue a left shift in thespectral distribution was consistently found. Forthe force dependency of the spectral distribution,however, inconsistent "ndings were reported, de-pending on the muscle under test, the force leveland the quantitative measure used for the descrip-tion of the spectral distribution. The most oftenused measures to characterize EMG power spectraldistribution are the Median Frequency (MF) ac-cording to Stulen and De Luca (1981) and theMean Power Frequency (MPF) as de"ned byKwatny et al. (1970). Measurements have been per-formed to study the MPF and MF as a functionof the muscular force and torque level duringboth static and dynamic contractions (e.g. Komiand Viitasalo, 1976; Viitasalo and Komi,1977,1978; Petrofsky, 1980; Petrofsky and Lind,1980; Petrofsky et al., 1982; Hagberg and Ericson,1982; Broman et al., 1985; Moritani and Muro,1987; Nagata et al., 1990; Bilodeau et al.,1990,1991,1992a,b). A collation of literature dataregarding the relationship between the force andboth spectral measures is provided for studies onextremity muscles in Fig. 2. A relatively uniformresult is achieved for MF. In most of the studies forvarious muscles and di!erent types of contractions(step contraction, ramp contraction) an increase in


Fig. 2. Collation of literature data: Median Frequency and Mean Power Frequency of electromyograms of di!erent extremity musclesas a function of the muscular force level expressed as the percentage of the force during maximum voluntary contraction (MVC).

MF is observed with increasing force up to forces ofmore than 60% of the maximum voluntary con-traction force (MVC). For MPF also an increase isshown in the majority of the studies for force levelsup to 30% and a levelling o! at higher forces. Insome cases even a decrease in MPF over the totalforce range was observed. In the following, due toits more uniform behavior the use of MF was

preferred in the JASA method for the characteriza-tion of the dependency of the EMG spectrum onthe force level.

The principle relationships between the EMGamplitude and spectrum, on the one hand, and thefatigue state and force production, on the otherhand, are illustrated schematically in Fig. 3.The amplitude is assumed to be quanti"ed by the


Fig. 3. Schematic representation of the method for the JointAnalysis of EMG Spectrum and Amplitude (JASA): Time-re-lated changes in the Electrical Activity (EA) and the MedianFrequency (MF) are considered jointly in order to di!erentiatebetween various EMG-change causations.

Electrical Activity (EA) which is derived from theraw EMG by recti"cation and low-pass "ltering.Spectral distribution is characterized by MF.A temporal increase in EA } illustrated in theright-hand half-plane of the xy diagram in Fig.3 }may result from an increase in the force produc-tion over time or from fatigue. A temporal decrease,shown in the left-hand half-plane, may be e!ectedby force decrease or recovery from previous fatigue.If such a change in the amplitude is consideredsolely, the discrimination between a change in thefatigue state or in the force production is obviouslynot possible. A similar problem arises with respectto the spectrum: A left shift in the power spectraldensity indicated by a negative temporal change inMF } see the lower half-plane in Fig. 3 } may bereceived from fatigue or, according to Fig. 2, fromdecrease in force. Analogously, in the upper halfplane an increase in MF may result from forceincrease or recovery. If both characteristics of theEMG are considered jointly, more reliable con-clusions can be drawn than from the use of oneof the characteristics only. Four cases can be dis-tinguished: (i) For EMG recordings for which asimultaneous increase in EA and MF over time isobserved, the concordant statement drawn from

the change in amplitude and spectrum is an in-crease in muscle force (upper right-hand quadrantin the xy diagram in Fig. 3). (ii) An increase in EAwhich occurs together with a decrease in MF canbe regarded as the result of muscle fatigue (lowerright-hand quadrant). (iii) Similarly a decrease inEA and an increase in MF points to recovery fromprevious muscle fatigue (upper left-hand quadrant).(iv) A decrease in EA accompanied by a decrease inMF can be regarded as a sign of muscle forcedecrease (lower left-hand quadrant).

When applying the JASA procedure describedabove, knowledge about the temporal behaviour ofthe EMG amplitude and spectrum or measuresthereof is needed. In previous studies (e.g. Lutt-mann and JaK ger, 1992; Luttmann et al.,1989,1996a,b) such information were obtained fromlong-term EMG recordings by calculating meanvalues of EA and MF for succeeding short periodsof time (e.g. 5 s or 10 s). This procedure results intime series for both characteristics with the respect-ive sampling interval. EA and MF time series weresummarized by regression analyses and the regres-sion coe$cients for EA and MF were regarded asquantitative indicators of the temporal change inthe EMG amplitude and spectrum.

3. Examples for occupational fatigue analyses

3.1. EMG recording and evaluation

Electromyographical "eld studies using theaforementioned techniques for selecting referenceactivities as well as the JASA method were per-formed in various occupational settings during to-tal shifts (e.g. manual materials handling ina weaving mill, price-recording at scanner check-outs, surgical work using endoscopic operationtechniques). In all cases surface electromyogramsfrom various muscles were recorded using a telem-etric procedure. For this purpose electrodes werea$xed to the skin by self-adhesive rings. After pre-ampli"cation up to 5 EMG signals were transmit-ted via a telemetric system and recorded on ananalog magnetic tape. In data evaluation from theraw EMG the Electrical Activity (EA) was formedby recti"cation and continuous averaging over


Fig. 4. Two sections of an original recording of the electromyograms (EMG) and Electrical Activity (EA) of the biceps and "nger #exorsduring the handling of bobbins in a weaving mill, with a time lapse of about half an hour between the sections; black bars denote sectionswith lifting of bobbins.

a time window of particular length (between 140and 400 ms); the EMG spectra were calculatedfrom the raw EMG by Fourier transformation, andthe Median Frequency was determined in order toquantify the spectral distribution.

Parallel to the EMGs the action code describingthe occupational activity of the person under test(cf. Section 2.3) was recorded on the same tape andused in data evaluation for the selection of referenceactivities representing EMG sections which are as-sociated with similar muscular force production.

3.2. Fatigue indication using reference activities

The "rst example for an occupational fatigueanalysis based on the evaluation of EMGs duringreference activities is taken from a study on personsmanipulating objects in a weaving mill. The task ofthe person under test was to produce warp beams.In the course of this activity bobbins with a mass of

about 10 kg had to be transferred from transportcontainers into a special rack arrangement, calledbeamer. Grasping and lifting of the bobbins re-quires high muscular forces of the arm and "nger#exors. Accordingly EMGs of the m. biceps brachiiand the m. #exor digitorum super"cialis of botharms were recorded and analysed. The activity ofthe person under test was documented during thework using the aforementioned coding system (seeFig. 1). During the study the following activitieswere encoded among others: lifting a bobbin,mounting a bobbin onto a peg in the beamer, andtying threads together.

For the purpose of fatigue analysis the liftingactivity was chosen as a reference activity and thetemporal behaviour of the EMG amplitude forsuch activity sections was followed up. The proced-ure will be explained using Fig. 4: Two short EMGsections, each lasting about 3 min, are shown. Thetime interval between the sections amounts to


Fig. 5. Mean Electrical Activity of various shoulder and armmuscles during scanning (i.e. price recording by moving anarticle over an optical scanner) and scanning rate for a totalworking day at a check-out station in a supermarket. Timedependencies are represented by regression lines (straight lines)for 5 work sections.

about half an hour. Each lifting of a bobbin isindicated by a black bar underneath the EMGrecordings. The point in time and the length of eachlifting period were derived from the correspondingaction code signal. Fig. 4 indicates that the highestEMG amplitudes occur during the lifting periods.Furthermore, it becomes evident from the EA re-cording (lowest trace of Fig. 4) that the peaks with-in the time curves are higher in the right sectionthan in the left one. This points to an increase in themyoelectrical activity in the course of the work.The "nding shown here for short sections only, iscon"rmed by a detailed analysis of all of the peakactivities for total shifts (Luttmann and JaK ger,1992). It is concluded that the observed increase inthe EMG peak amplitude during lifting indicatesthe occurence of muscular fatigue. The other pos-sible explanation for the increase in the EMG am-plitude, namely the increase in muscular force, isunlikely for this activity, since the mass and theform of the handled objects were unchanged andthe lifting technique as well as the body posture didnot vary systematically.

In Fig. 5 another example for the fatigue analysisbased on the determination of the EMG amplitudefor reference activities is shown. Data from a studyof persons working on scanner check-outs ina supermarket (Luttmann et al., 1989) are provided.EMGs were derived from various shoulder muscles(m. trapezius, m. deltoideus) and arm muscles (m.biceps brachii, m. extensor carpi ulnaris) and re-corded using the aforementioned telemetric system.Muscles from the left-hand side of the body werechosen, since, due to the construction of the cash-desk, almost all articles were moved using the lefthand. The current activity of the subject was re-corded by producing an action code according toFig. 1. Additionally, an electrical scanner signal wasrecorded which was triggered by each movement ofan article across the scanner in order to count thenumber of price recordings per unit of time. Incontrast to the example shown in Fig. 4, whereshort sections of the EMG recordings are provided,here data for a whole working day are demon-strated.

In the upper traces of Fig. 5 each point refersto the average Electrical Activity in a time sec-tion relating to the action `scanninga (i.e. price

recording by moving articles over the scanner). Inthe lowest trace, as a measure of the working speed,the number of scanned articles per minute is re-corded. The time courses demonstrate that thework over the day is divided by breaks (see theblack bars) into 5 activity sections. Within the sec-tions the EA data and the number of articles aresummarized by regression lines. A signi"cant in-crease in the EA is found for at least one of themuscles for four of the "ve working sections. Bycontrast, the regression lines for the number ofarticles do not increase or decrease signi"cantlyover time; so the mean workload within the sec-tions can be interpreted as not changing. It is there-fore concluded that the measured increase in themean EA is not the result of a change in load but anindication of muscle fatigue. The results shown in


Fig. 5 for one subject during a total work shift werecon"rmed by further measurements on 4 personsduring total working days (Luttmann et al., 1989).

3.3. Fatigue indication using the JASA method

The method described in the previous sectioncan only be applied, if an identi"cation of referenceactivities comprising similar workload is possible.In a more general approach the newly developedJASA method for the joint analysis of EMG spec-trum and amplitude is used in which, in addition tothe change in the EMG amplitude, the change inthe spectral distribution is considered.

An application of the method will be demon-strated referring to an example of an occupationalEMG "eld study performed on surgeons in anoperation theatre (Luttmann et al., 1996a). TheEMGs from several muscles in the shoulder-armregion (right and left m. trapezius, right m.deltoideus) and the back (left m. erector spinae)were derived from surgeons during endoscopic op-erations in urology. In such operations a so-calledresectoscope is introduced into the body throughthe urethra. The resectoscope mainly consists ofa rod-shaped endoscope used for the visual inspec-tion of the operation area and a wire loop suppliedby a high frequency current for the dissection orcoagulation of the tissue. The "rst part of the studywas performed when the surgeons used the so-called `direct endoscopya. In this method, the sur-geons looked `directlya into the urethra and thebladder through the endoscope. This method re-quires that both hands of the surgeons simulta-neously grasp the instrument and that one eye ispermanently in contact with the aperture of theendoscope. When positioning the instrument it isoften necessary to steeply incline the trunk and toremain in this posture for a long time. A typicalbody posture during an operation using direct en-doscopy is shown in the upper photograph in Fig. 6.

Data from the "rst part of the study were com-pared with the results of a second part which wasperformed after an ergonomic intervention. In thisa video system was introduced enabling the sur-geon to apply the `monitor endoscopya. For thispurpose a camera was mounted on the top of theendoscope and the operation area could be visually

inspected using a monitor. The monitor could beadjusted according to the anthropometry of thesurgeon and the patient. Furthermore, a chair witharm and back rests was used which allows to sup-port the back and the elbows of the surgeons fromtime to time in order to unburden the muscles. Inthe lower photograph in Fig. 6, the newly intro-duced equipment (HoK cker et al., 1995) including themonitor hanging on the ceiling and being adjust-able in the horizontal and vertical directions as wellas the newly developed chair (SoK keland et al., 1995)is shown. Using this device the operation can beperformed with the trunk held in an almost uprightposition.

In the "rst part of the investigation, the EMGstudies were performed on four surgeons during 14endoscopic operations lasting between 18 and83 min. The second part was carried out more thanone year after the ergonomic redesign of the opera-tion equipment when the surgeons were accus-tomed to the new arrangement including the videotechnique. It was performed on 5 surgeons during12 operations lasting between 18 and 80 min.

EMG amplitude was quanti"ed by determiningthe Electrical Activity. Spectral distribution wasanalysed by performing Fourier transformationand calculating the Median Frequency. The tem-poral behaviour of both characteristics was evalu-ated as described before in Section 2.4 bycomputing the means for EA and MF for sectionsof 10 or 5 s length, respectively. For the resultingEA and MF time series a linear regression analysiswas performed. The slopes of the regression linesfor EA and MF are used in the further analysis ascharacteristics of the temporal behaviour of theEMG. According to the principle of the JASA ap-proach (see Fig. 3) changes in the EMG were at-tached to one of the categories fatigue, recovery,force increase, or force decrease.

Results from the "rst part of the study during theapplication of direct endoscopy are provided inTable 1 for a total of 45 EMG recordings during 14operations. (Data from 11 recordings could not beconsidered in data evaluation owing to distur-bances induced by high frequency currents used forthe dissection and coagulation of the tissue.) Itreveals that fatigue occurs in the right trapezius inthe majority of the operations (11 to 14). With


Fig. 6. Left-hand part: Photographs of typical work situations during an urological operation applying direct endoscopy beforeergonomic redesign (above) and using monitor endoscopy after redesign (below). Right-hand part: Corresponding joint analysis of EMGspectrum and amplitude (JASA) for EMG recordings from the right m. trapezius of the surgeons.

Table 1Time-related changes in the amplitude and spectrum of elctromyograms of various muscles derived from surgeons during urologicaloperations applying `direct endoscopya!

Number of operations with simultaneous

EA increase andMF decrease

EA decrease andMF increase

EA increase andMF increase

EA decrease andMF decrease

Interpretation Fatigue Recovery Force increase Force decrease

M. trapezius, right 11 0 1 2M. trapezius, left 4 1 3 1M. deltoideus, right 4 2 3 0M. Erector spinae, left 4 1 4 4

!EA: Electrical Activity; MF: Median Frequency.

respect to the other muscles fatigue was identi"edfor 4 operations in all cases. It may concluded thatthe trapezius muscle forms a bottleneck for the

performance of this activity. Signs of recoverywere found altogether in 4 cases, only. The numberof operations for which a change in the force


production was identi"ed depends on the muscle, itvaries between 4 (in case of the m. erector spinae)and 0 (m. deltoideus). It is concluded that duringdirect endoscopy fatigue occurred in the majorityof the operations, the right trapezius muscle beingparticularly a!ected.

In Fig. 6 the in#uence of the ergonomic interva-tion on body posture and muscle fatigue is shown.In the left-hand part photographical representionsof the typical pustures of the persons under testbefore and after redesign are shown. Before re-design, direct endoscopy is applied and a steeplateral bending of the surgeon can be seen; afterredesign, when using the monitor method, the op-eration can be performed in an almost uprightposition. In the right-hand part of Fig. 6 the com-parison of EMG "ndings from both parts of thestudy is provided. Data are focussed on the righttrapezius muscle since this muscle was identi"ed toform a bottleneck for the execution of the opera-tion. In accordance with the schematical descrip-tion of the JASA method presented in Fig. 3, thetime-related change in EA is plotted on the x axisand the change in MF on the y axis. For eachoperation a pair of values was formed using the EAand MF regression coe$cients and graphically pre-sented in form of a dot in the xy diagram providedin the right-hand part of Fig. 6. Before redesign(upper diagram) about 80% of the EMGs are char-acterized by an increase in EA over time and a de-crease in MF. The corresponding dots are locatedin the `fatigue quadranta (lower right-hand quad-rant in the diagram). After redesign (lower diagram)this amount was lowered to about 45%. It is con-cluded that for the muscle a!ected most, i.e. theright trapezius muscle, fatigue is reduced by theergonomic intervention but not completely avoided.

4. Discussion

4.1. Methodological restrictions

EMG amplitude and spectrum depend not onlyon the fatigue state but also on the produced mus-cular force. Nevertheless, the separate use of theincrease in the EMG amplitude or the left shift inthe EMG spectrum for the indication of fatigue is

possible, if the force production is similar for allEMG sections which are included in the analysis ofthe EMG time course. This has been proven by theuse of test contractions (HaK gg et al., 1987). Theapplication of test contractions in occupationalstudies reveals some problems, however, since thebody postures which are adopted during the perfor-mance of the test contraction have to be repro-duced and controlled with high accuracy; otherwisefatigue-induced changes in the EMG might becovered by changes in the EMG caused by alter-ations in muscle geometry due to postural vari-ations and changed muscular force production.Furthermore, in many real activities, such as theexecution of the surgical work shown in the pre-vious section, it is not possible to in#uence thecontent of the work by incorporating such testcontractions.

Another possibility for selecting sections withsimilar force production is to use `reference con-tractionsa which are a natural part of the normalwork of the person under test. Here also, strictlyspeaking, only sections should be compared forwhich an identical force production is attained.Under real occupational conditions this demandcan be ful"lled with some restrictions only. In "eldsituations the same activity is normally performedsomewhat diversely, e.g. with respect to the bodyposture, working velocity, and working technique.This results in a certain scattering of the EMG datafor the same activity. Nevertheless, reliable in-formations regarding the EMG time course can beobtained and signi"cant changes in EMG can befound, if the selected reference activities form anessential part of the work and are, in consequence,performed with a high degree of repetition. Thenthe number of EMG samples usable for the timecourse analysis is high and statistical methods suchas regression analyses can be used in order to provethe signi"cance of EMG changes as shown in theexample provided in Fig. 5.

The indication of muscular fatigue is enhancedby applying the JASA method since changes in theEMG due to a change in the force production canbe separated from fatigue-induced changes. In theJASA approach, muscular fatigue is assumed incases, only, where an increase in EA as well asa decrease in MF over time was observed. In the


graphical representation of the JASA diagram(Fig. 6) the corresponding dots are positioned inthe lower right-hand quadrant. It should benoted, however, that the locations of the dots with-in this quadrant scatter within a wide range. Somedots are located near the x or y axis, respectively,others are lying in an `intermediate "elda betweenthe axes. For dots near one of the axes a distinctchange is observed in one of the characteristics EAor MF with only a slight change in the otherparameter; for dots in the intermediate "eld bothparameters are changing considerably over time.A systematic analysis of the causation for the di!er-ent locations of the dots using the data shown inFig. 6 is not advisable, since the respective EMGswere recorded under "eld conditions and somesupposed in#uencing factors like body posture orbody movements were not controlled with su$-cient accuracy. Therefore, corresponding laborat-ory studies with fatiguing muscle contractions ofconstant force were performed. In such studies anin#uence of the size of the muscle and the indi-vidual force generating capacity of the muscle un-der the test on the individual reaction in theparameters EA and MF was observed (Luttmannet al., 1997,1998). Accordingly, the individual max-imum force maybe partly responsible for the indi-vidual reaction. For further interpretationsadditional studies are necessary.

4.2. Physiological implications

The basis for the JASA method are experi-mentally proved relationships between EMGamplitude and spectrum and its measures EAand MF, on the one hand, and force and fatigue,on the other hand. Nevertheless, the appointmentof a change in the phenomenological parametersEA and MF to the physiological causes is di$cult,due to the complexity of the physiological pro-cesses.

An increase in muscular force is accompanied byan increase in EA. This mainly results from anincrease in muscular activation, namely, of thenumber of action potentials per unit of time. It isperformed by recruiting additional muscle "bresand/or by an increase in the "ring rate (`rate cod-inga). The relative importance of the two neural

mechanisms was described by Stein (1974). Inmuscle fatigue also an increase in EA is found. Inthis case two di!erent factors contribute to the EAincrease: (i) An increase in the activation of themuscle by recruitment and/or rate coding and (ii)a slowing in the propagation of the action poten-tials along the muscle "bres. Re (i) The fatigue-induced increase in the activation of the muscleoccurs, since, as a result of fatigue, the force peraction potential drops. (For review of the underly-ing physiological mechanisms see Luttmann(1996).) In order to maintain a constant force eitheradditional motor units are recruited or the actionpotential "ring rate is increased. Re (ii) Slowing ofthe action potential propagation velocity resultsfrom changes in the ional composition of the intra-and extra-cellular #uids in the muscle "bres and, inparticular, in changes in the potassium distribution(KoK ssler et al., 1990; Sj+gaard, 1990). Such a de-crease in the action potential propagation is re#ec-ted in a change in the EMG signal, in particular ina `broadeninga of the bipolar action potentialsmeasured on the skin surface. Both changes, theincrease in the number of action potentials andtheir broadening, results in an increase in EA, sinceEA is determined by integrating the EMG andtherefore, it depends upon the area below the EMGtime functions. Another often applied integrativemeasure of the EMG amplitude is the Root MeanSquare (RMS); its dependency on muscular forceand fatigue is similar to EA.

In summary, integrative measures of the EMGamplitude such as EA and RMS are suitable toillustrate changes in the EMG signal caused bymuscular fatigue; they are, however, not appropri-ate to solve the reverse problem, namely, to decidewhether an increase in the EMG amplitude resultsfrom fatigue.

Another measure for the description of the EMGamplitude, the Muscular Activity (ACT), was de-"ned some years ago by Spaepen et al. (1987). ACTis derived from the EMG not by integration, but bydi!erentiation. It increases with increasing muscu-lar activation and, in contrast to EA and RMS, itdecreases with decreasing propagation velocity(Hermans, 1996). The simultaneous use of an in-tegrative measure (EA or RMS) and a di!erenti-ating one (ACT) allows to distinguish between


force-related and fatigue-induced EMG changes.Based on this idea, a combination of both types ofmeasures of the EMG amplitude, the ratio betweenRMS and ACT, was introduced recently asan indicator of muscular fatigue (Spaepen andHermans, 1996).

An increase in muscular force is not only accom-panied by an increase in the EMG amplitude butalso by an increase in MF (see Fig. 2) which re#ectsa spectral shift to higher frequencies and is causedmainly by an increase in the number of actionpotentials per unit of time. Such behaviour of MFin relation to the "ring rate is con"rmed by modelstudies of Hermens et al. (1992). The change in thespectral distribution occurring during fatiguingcontractions is in#uenced by counteracting e!ects:(i) A shift to higher frequencies due to the aforemen-tioned increase in the number of action potentialsand (ii) a shift towards lower values. In the litera-ture this left shift in the spectrum has been dis-cussed intensely for many years. (For review seeBasmajian and De Luca (1985); De Luca (1985);HaK gg (1992); HaK gg and Kadefors (1996).) Two maininterpretations have emerged: reduction in actionpotential conduction velocity and synchronizationor grouping of motor unit "ring. The reduction inconduction velocity is currently the most widelyacknowledged interpretation of the fatigue-inducedspectral change even if the decrement in spectralparameters is often greater than the decrement inconduction velocity (Broman et al., 1985; Merlettiet al., 1992). However, there is some evidence thatthis explanation is only partially valid, in particularfor low- and moderate-level endurance contrac-tions (Krogh-Lund, 1993; Ca$er et al., 1993). Thenet spectral change normally observed during fa-tiguing contractions is to the left; it is concludedthat the spectral shift towards lower values exceedsthe shift to higher ones.

One of the main interests in the fatigue analysisin occupational health and ergonomics is to de"nean indicator of the loss in the force generatingcapacity of the muscle under test. This informationcan then be applied for the justi"cation of measuresfor work design like redesign of the workplaceaccording to ergonomic principles or the introduc-tion of a work-rest regimen which prevents immod-erate fatigue. Another aim of fatigue analysis deals

with musculoskeletal disorders and functional in-su$ciency. Static activity can be assumed to serveas a risk factor for the development of muscularpain symptoms (Westgaard et al., 1996), in particu-lar for the trapezius muscle. In this muscle highstatic activity was observed, even if objects of lowweight were handled (Strasser and MuK ller, 1999).The prevailing hypothesis for the occurence ofshoulder myalgia is based on observations thatlow-threshold motor units are active throughoutthe contraction, until total relaxation (Kadeforset al., 1999). Therefore, the reduction of static activ-ity by ergonomic work design and the introductionof breaks allowing muscular relaxation are as-sumed to be e!ective measures for the prevention ofmyalgia symptoms.

The aforementioned short compilation of the re-lationships between the EMG changes and thephysiological processes indicates that all com-monly used fatigue measures characterize changesin the myoelectrical activity, which coincide withfatigue in a complex way, but do not depend dir-ectly on mechanical characteristics. This elucidatesthe necessity of further development of an adequatemeasure of muscular fatigue which is preferablyderived from surface EMGs and which dependsdirectly on the interrelationship between the elec-trical EMG signal and the mechanical force devel-opment. A possible way for the derivation of sucha measure is to determine an electrical equivalent ofthe `elementary forcea, i.e., the force whichis produced by a single action potential, and itsfatigue-induced reduction.

4.3. Conclusions regarding work design

An activity in which the working musculature isfatigued can only be performed for a limited periodof time. After this time has elapsed, the musculatureis exhausted and the needed force cannot be pro-duced any longer. Another implication of fatigue isa loss in the required accuracy of movements. Froma practical point of view, indication of muscularfatigue is needed in occupational studies in order toquantify the time for performing a certain activitywithout reaching the state of immoderate fatigueand disabling the person to continue the activitywith the necessary precision. Such queries can be


studied using EMG "ndings, if the connectionbetween fatigue-induced changes in the EMGand the endurance time of the muscle is takeninto consideration. In the frequency domain,Hagberg (1981) has studied this connection byanalysing the relationship between the temporalchange in MPF and the endurance time. In theamplitude domain, the increase in EA caused byfatigue and the resulting endurance time was inves-tigated by Laurig (1974,1975) using a fairly largenumber of experiments on di!erent muscle groups.It reveals that a steep increase in the EA is connec-ted with a small endurance time whereas a #atincrease in EA is related to a large endurance time.The data were later extended by some furthermeasurements (Laurig et al., 1987) and summarizedusing the regression formula y"101.622~0.745 -0' x

with y"endurance time in min and x"increasein Electrical Activity in percent per min (Luttmann,1996).

For practical applications such relationships canbe used to predict the endurance time for the mea-sured EMG change. Investigations of this type wereperformed within the aforementioned studies onpersons executing manual handling tasks in aweaving mill, during cash-desk activities and forsurgeons performing endoscopic operations inurology. In all cases the periods of time for uninter-rupted working sections were compared with thepredicted endurance times, and if necessary,measures for the reduction in muscular strain andfatigue by introducing a new ergonomic work de-sign or a physiologically justi"ed work-rest regi-men with shorter working periods could be derivedand were recommended.

References

BalleH , W., Smolka, R., Luttmann, A., 1982. Aufbau und Er-pobung eines Me{fahrzeugs zur Datenerfassung in arbeits-physiologischen Felduntersuchungen. Zentralblatt fuK rArbeitsmedizin 32, 214}218.

Basmajian, J.V., De Luca, C., 1985. Muscles Alive - Their Func-tions Revealed by Electromyography 5th Edition. Williamsand Wilkins, Baltimore.

Bilodeau, M., Arsenault, A.B., Gravel, D., Bourbonnais, D.,1990. The in#uence of an increase in the level of force on theEMG power spectrum of elbow extensors. European Journalof Applied Physiology 61, 461}466.

Bilodeau, M., Arsenault, A.B., Gravel, D., Bourbonnais, D.,1991. EMG power spectrum of elbow extensors during rampand step isometric contractions. European Journal of Ap-plied Physiology 63, 24}28.

Bilodeau, M., Arsenault, A.B., Gravel, D., Bourbonnais, D.,1992a. In#uence of gender on the EMG power spectrumduring an increasing force level. Journal of Electromyogra-phy and Kinesiology 2 (3), 121}129.

Bilodeau, M., Arsenault, A.B., Gravel, D., Bourbonnais, D.,1992b. Time and frequency analysis of EMG signals ofhomologous elbow #exors and extensors. Medical & Bio-logical Engineering & Computing 30, 640}644.

Broman, H., Bilotto, G., de Luca, C.J., 1985. Myoelectric signalconduction velocity and spectral parameters: in#uence offorce and time. Journal of Applied Physiology 58 (5),1428}1437.

Ca$er, G., Heinecke, D., Hinterthan, R., 1993. Surface EMGand load level during long-lasting static contractions of lowintensity. International Journal of Industrial Ergonomics 12,77}83.

Cobb, S., Forbes, A., 1923. Electromyographic studies of muscu-lar fatigue in man. Americal Journal of Physiology 65,234}251.

De Luca, C., 1985. Myoelectric manifestations of localizedmuscle fatigue in humans. CRC Critical Reviews in Biomedi-cal Engineering 11, 251}279.

HaK gg, G.M., SuurkuK la, J., Liew, M., 1987. A worksite method forshoulder muscle fatigue measurements using EMG, test con-tractions and zero crossing technique. Ergonomics 30,1541}1551.

HaK gg, G.M., 1992. Interpretation of EMG spectral alterationsand alteration indexes at sustained contraction. Journal ofApplied Physiology 73, 1211}1217.

HaK gg, G.M., Kadefors, R., 1996. EMG alterations at sustainedcontractions with special emphasis on applications in ergo-nomics. In: Kumar, S., Mital, A. (Eds.), Electromyography inErgonomics. Taylor and Francis, London, pp. 163}181.

Hagberg, M., 1981. Muscular endurance and surface elec-tromyogram in isometric and dynamic exercise. Journal ofApplied Physiology 51, 1}7.

Hagberg, M., Ericson, B.-E., 1982. Myoelectric power spectrumdependence on muscular contraction level of elbow #exors.European Journal of Applied Physiology 48, 147}156.

Hermens, H.J., v. Bruggen, T.A.M., Baten, C.T.M., Rutten,W.L.C., Boom, H.B.K., 1992. The median frequency of thesurface EMG power spectrum in relation to motor unit"ring and action potential properties. Journal of Elec-tromyography and Kinesiology 2, 15}25.

Hermans, V., 1996. Changes in surface EMG signals of shoulderand neck muscles during sustained e!ort. Ph.D. Thesis.Katholieke Universiteit Leuven.

HoK cker, F., Luttmann, A., SoK keland, J., Witkowski, M., 1995.Versorgungseinheiten fuK r endoskopische ArbeitsplaK tze.Urologe B 35, 265}266.

Kadefors, R., Forsman, M., ZoeH gas, B., Herberts, P., 1999. Re-cruitment of low threshold motor-units in the trapeziusmuscle in di!erent static positions. Ergonomics 42, 359}375.


KoK ssler, F., Ca$er, G., Lange, F., 1990. Probleme der Muskeler-muK dung - Beziehung zur Erregungsleitungsgeschwindigkeitund K`-Konzentration. Zeitschrift fuK r die gesamte Hygiene36, 354}356.

Komi, P.V., Viitasalo, J.H.T., 1976. Signal characteristics ofEMG at di!erent levels of muscle tension. Acta PhysiologicaScandinavia 96, 267}276.

Krogh-Lund, C., 1993. Myo-electric fatigue and force failurefrom submaximal static elbow #exion sustained to exhaus-tion. European Journal of Applied Physiology 67, 389}401.

Kumar, S., 1996. Electromyography in ergonomics. In: Kumar,S., Mital, A. (Eds.), Electromyography in Ergonomics.Taylor and Francis, London, pp. 1}50.

Kwatny, E., Thomas, D.H., Kwatny, H.G., 1970. An applicationof signal processing techniques to the study of myoelectrics5 ignals. IEEE Transactions on Biomedical Engineering 17,303}313.

Laurig, W., 1974. Beurteilung einseitig dynamischer Muskelar-beit. Beuth, Berlin.

Laurig, W., 1975. Methodological and physiological aspects ofelectromyographic investigations. In: Komi, P.V. (Ed.), Bi-omechanics V-A. University Park, Baltimore, pp. 219}230.

Laurig, W., Luttmann, A., JaK ger, M., 1987. Evaluation of strainin shop-#oor situations by means of electromyographic in-vestigations. In: Asfour, S.S. (Ed.), Trends in Ergonom-ics/Human Factors IV. Elsevier Science Publishers,Amsterdam, pp. 685}692.

Luttmann, A., 1996. Physiological basis and concepts ofelectromyography. In: Kumar, S., Mital, A. (Eds.), Electro-myography in Ergonomics. Taylor & Francis, London,pp. 51}95.

Luttmann, A., JaK ger, M., 1992. Reduction of muscular strain bywork design: Electromyographical "eld studies in a weavingmill. In: Kumar, S. (Ed.), Advances in Industrial Ergonomicsand Safety IV. Taylor & Francis, London, Washington,pp. 553}560.

Luttmann, A., JaK ger, M., 1998. Discrimination between fatigue-induced and force-related EMG changes in occupational"eld studies. In: Kumar, S. (Ed.), Advances in OccupationalErgonomics and Safety 2. IOS Press, Amsterdam, pp.218}221.

Luttmann, A., JaK ger, M., Laurig, W., 1988. Surface electromyo-graphy in work-physiological "eld studies for the analysis ofmuscular strain and fatigue. In: Wallinga, W., Boom, H.B.K.,de Vries, J. (Eds.), Electrophysiological Kinesiology. ElsevierScience Publishers, Amsterdam, pp. 301}304.

Luttmann, A., JaK ger, M., Laurig, W., 1989. Elektromyo-graphische Untersuchungen an KassenarbeitsplaK tzen mitScannern. Zeitschrift fuK r Arbeitswissenschaft 43 (15 NF),234}240.

Luttmann, A., JaK ger, M., SoK keland, J., Laurig, W., 1996a. Elec-tromyographical study on surgeons in urology, Part II: De-termination of muscular fatigue. Ergonomics 39, 298}313.

Luttmann, A., JaK ger, M., SoK keland, J., Laurig, W., 1996b. Jointanalysis of spectrum and amplitude (JASA) of electromyo-grams applied for the indication of muscular fatigue amongsurgeons in urology. In: Mital, A., Krueger, H., Kumar, S.,

Menozzi, M., Fernandez, J.E. (Eds.), Advances in Occupa-tional Ergonomics and Safety. Int. Soc. for OccupationalErgonomics and Safety, Cincinnati, pp. 523}528.

Luttmann, A., JaK ger, M., Witscher, K., Vorgerd, M., Tegentho!,M., 1997. Individual fatigue response in the joint analysis ofEMG spectrum and amplitude (JASA). In: Hermens, H.J.,HaK gg, G., Freriks, B. (Eds.), European Applications on Sur-face Electromyography. Roessingh Research and Develop-ment, Enschede, pp. 138}146.

Luttmann, A., JaK ger, M., Witscher, K., Vorgerd, M., Tegentho!,M., 1998. Fatigue-induced EMG changes during low-levelisometric contractions. In: Hermens, H.J., Rau, G., Dissel-horst-Klug, C., Freriks, B. (Eds.), Surface Electromyography:Application Areas and Parameters - Proceedings of theThird General SENIAM Workshop. Roessingh Researchand Development, Enschede, pp. 141}150.

Merletti, R., Kna#itz, M., De Luca, C.J., 1992. Electricallyevoked myoelectric signals. CRC Critical Reviews in Bio-medical Engineering 19, 293}340.

Moritani, T., Muro, M., 1987. Motor unit activity and surfaceelectromyogram power spectrum during increasing force ofcontraction. European Journal of Applied Physiology 56,260}265.

Nagata, S., Arsenault, A.B., Gagnon, D., Smyth, G., Mathieu,P.-A., 1990. EMG power spectrum as a measure of muscularfatigue at di!erent levels of contraction. Medical & Bio-logical Engineering & Computing 28, 374}378.

Petrofsky, J.S., 1980. Computer analysis of the surface EMGduring isometric exercise. Computers in Biology and Medi-cine 10, 83}95.

Petrofsky, J.S., Lind, A.R., 1980. Frequency analysis of thesurface electromyogram during sustained isometric con-tractions. European Journal of Applied Physiology 43,173}182.

Petrofsky, J.S., Glaser, R.M., Phillips, C.A., Lind, A.R., Williams,C., 1982. Evaluation of the amplitude and frequency compo-nents of the surface EMG as in index of muscle fatigue.Ergonomics 25 (3), 213}223.

Piper, H., 1909. UG ber die ErmuK dung bei willkuK rlichen Muskel-kontraktionen. Archiv fuK r Anatomie und Physiologie491}498.

Piper, H., 1912. Elektrophysiologie menschlicher Muskeln.Springer, Berlin.

Sj+gaard, G., 1990. Exercise-induced muscle fatigue: The signi"-cance of potassium. Acta Physiologica Scandinavia 140(Suppl. 593).

SoK keland, J., Luttmann, A., Seidel-Fabian, B., 1995. Ergonomieam endoskopischen Arbeitsplatz. Urologe B 35, 428}429.

Spaepen, A.J., Baumann, W., Meas, H., 1987. Relation betweenmechanical load and EMG-activity of selected muscles of thetrunk under isometric conditions. In: Bergmann, G., KoK lbel,R., Rohlmann, A. (Eds.), Biomechanics: Basic and AppliedResearch. Martinus Nijho! Publishers, Dordrecht, pp.473}478.

Spaepen, A., Hermans, V., 1996. EMG in occupational settings:Measurement of muscle load or muscle fatigue. In: Mital, A.,Krueger, H., Kumar, S., Menozzi, M., Fernandez, J.E. (Eds.),


Advances in Occupational Ergonomics and Safety. Interna-tional Society for Occupational Ergonomics and Safety, Cin-cinnati, pp. 557}560.

Stein, R.B., 1974. Peripheral control of movement. PhysiologicalReviews 54, 215}243.

Strasser, H., MuK ller, K.-W., 1999. Favorable movements of thehand-arm system in the horizontal plane assessed by elec-tromyographic investigations and subjective rating. Interna-tional Journal of Industrial Ergonomics 23, 339}347.

Stulen, F.B., De Luca, C.J., 1981. Frequency parameters of themyoelectric signal as a measure of muscle conduction velocity.IEEE Transactions on Biomedical Engineering 28, 515}523.

Viitasalo, J.H.T., Komi, P.V., 1977. Signal characteristics ofEMG during fatigue. European Journal of Applied Physi-ology 37, 111}121.

Viitasalo, J.H.T., Komi, P.V., 1978. Interrelationships of EMGsignal characteristics at di!erent levels of muscle tension andduring fatigue. Electromyography and Clinical Neuro-physiology 18, 167}178.

Westgaard, R.H., Jansen, T., Jensen, C., 1996. EMG of neck andshoulder muscles: the relationship between muscle activityand muscle pain in occupational settings. In: Kumar, S.,Mital, A. (Eds.), Electromyography in Ergonomics. Taylorand Francis, London, pp. 227}258.


luttmann 2000 (artigo jasa)

Documents