表面肌电信号处理

内容需要下载文档才能查看

Downloaded from http://wendang.chazidian.com on May 31, 2010

Signal processing of the surface electromyogramto gain insight into neuromuscular physiology

J. Stefan Karlsson, Karin Roeleveld, Christer Grönlund, Andreas Holtermann andNils ÖstlundPhil. Trans. R. Soc. A 2009 367, 337-356doi: 10.1098/rsta.2008.0214

References

Rapid response

Subject collectionsThis article cites 51 articles, 8 of which can be accessed free http://wendang.chazidian.com/content/367/1887/337.full.html#ref-list-1Respond to this article http://wendang.chazidian.com/letters/submit/roypta;367/1887/337Articles on similar topics can be found in the followingcollections biomedical engineering (174 articles)

Receive free email alerts when new articles cite this article - sign upin the box at the top right-hand corner of the article or click here Email alerting service

To subscribe to Phil. Trans. R. Soc. A go to: http://wendang.chazidian.com/subscriptions

This journal is © 2009 The Royal Society

Downloaded from http://wendang.chazidian.com on May 31, 2010

Phil.Trans.R.Soc.A(2009)367,337–356

doi:10.1098/rsta.2008.0214

Publishedonline30October2008

Signalprocessingofthesurfaceelectromyogramtogaininsightintoneuromuscularphysiology

¨NLUND1,BYJ.STEFANKARLSSON1,*,KARINROELEVELD2,CHRISTERGRO¨STLUND1ANDREASHOLTERMANN2ANDNILSO

DepartmentofBiomedicalEngineering&Informatics,UniversityHospital,

90185Umea?,Sweden2HumanMovementSciencesProgramme,NorwegianUniversityofScienceand

Technology,DragvollIdrettssenter,7491Trondheim,Norway

Asurfaceelectromyogram(sEMG)containsinformationaboutphysiologicalandmorphologicalcharacteristicsoftheactivemuscleanditsneuralstrategies.Becausetheelectrodesaresituatedontheskinabovethemuscle,thesEMGisaneasilyobtainablesourceofinformation.However,differentcombinationsofphysiologicalandmorphologicalcharacteristicscanleadtosimilarsEMGsignalsandsEMGrecordingscontainnoiseandotherartefacts.Therefore,manysEMGsignalprocessingmethodshavebeendevelopedandappliedtoallowinsightintoneuromuscularphysiology.ThispapergivesanoverviewofimportantadvancesinthedevelopmentandapplicationsofsEMGsignalprocessingmethods,includingspectralestimation,higherorderstatisticsandspatio-temporalprocessing.Thesemethodsprovideinformationaboutmuscleactivationdynamicsandmusclefatigue,aswellascharacteristicsandcontrolofsinglemotorunits(conductionvelocity,?ringrate,amplitudedistributionandsynchronization).

Keywords:electromyography;multi-channel;spatio-temporalprocessing;

neuromuscularphysiology;musclefunction;musclecontrol1

1.Introduction

Surfaceelectromyographyisacommonlyappliedtechniquetoevaluatephysiologicalandmorphologicalcharacteristicsofmusclesandtheirneuralstrategies.Itinvolvesregistrationofelectricalpotentialsfromactivemuscle?bresbyelectrodesplacedontheskinabovethemuscle.Moreprecisely,therecordedsignalorthesurfaceelectromyogram(sEMG)consistsoftheweightedsumoftheelectricalcontributionsofactivemotorunits(MUs).TheMUisthesmallestfunctionalunitofmovement,andconsistsofamotorneuronandthegroupofmuscle?bresitinnervates.TheelectricalrepresentationofactiveMUs,andthereforethesEMG,containsinformationaboutthecharacteristicsandphysiologyoftheactiveMUs(e.g.MUsize,activationand?ringpattern,andmuscle?bretype,sizeandstate).Inaddition,thesEMGisrelatedtoforceproductionandcanbeusedtogaininsightintowhichsituationsMUs

are*Authorforcorrespondence(stefan.karlsson@vll.se).

Onecontributionof13toaThemeIssue‘Signalprocessinginvitalrhythmsandsigns’.

337Thisjournalisq2008TheRoyalSociety

Downloaded from http://wendang.chazidian.com on May 31, 2010

338J.S.Karlssonetal.

activated,thusenlighteningtheneuralstrategy.Furthermore,sincedetectionofthissignalinvolveselectrodesplacedontheskinabovethemuscle,itbarelyinterfereswithnormalmuscleactivityandbodymovements.However,thislargeamountofinformationinthesignalalsocausesmultiplesolutionstotheinverseproblemofdeterminingsingleneuromuscularproperties.Inaddition,therecordedsignalincludesnoise,interferenceandotherartefacts.ThesEMGdependsontheelectrodes,thetissuebetweentheelectrodesandthemuscle,andtheacquisitionequipment.ToallowinsightintoneuromuscularphysiologyfromthesEMG,threeindispensableapproachesareused:standardizationofdatacollection(Hermensetal.1999);investigationoftherelationshipbetweenthesignalandphysiology(e.g.Roeleveld&Stegeman2002);anddevelopmentandapplicationofsignalprocessingmethods.

Thispapergivesanoverviewofimportantrecentadvancesofdevelopedandappliedsignalprocessingmethodsinsurfaceelectromyographytogaininsightintoneuromuscularphysiology.Foramoreextensiveoverviewofmethodsappliedinsurfaceelectromyography,thereaderisreferredtoBasmajian&DeLuca(1985)andMerletti&Parker(2004).AfteranintroductionabouttheoriginofthesEMG(§2),apresentationofsomepre-processingmethods(§3),signalprocessingmethodsfortheestimationofmuscleactivity(especiallydynamicsofactivation,§4),thesEMGspectrumanditsdescriptors(§5),muscle?breconductionvelocity(CV;§6),singleMUactivity(decomposition;§7),MUsynchronization(§8)andlocalizationoftheinnervationzone(IZ;§9)arepresented.Ineachsection,?rst,thephysiologicalproblemorgoalisde?ned,thentheclassicalsolutionsaregivenandthendetailedexamplesofrecentmethods,mostlydevelopedbyourgroup,arepresented.Wherepossible,?rst,signalprocessingmethodsfortraditionalsingle-channelsEMGsignalsarepresentedfollowedbymethodsformulti-channelorhigh-densitysEMGsignals.Inmulti-channelsurfaceelectromyography,multipleelectrodesareplacedovertheskinofasinglemuscle,whilehigh-densitysurfaceelectromyographyisreferredtoasaspeci?ccaseofmulti-channelsurfaceelectromyographyinwhichrelativelysmallelectrodeswithsmallinter-electrodedistancesareapplied.

2.Thesurfaceelectromyogram

MUactionpotentials(MUAPs)arethemostfundamentalelementsofthesEMG.TheMUAPisthesummationofsingle-?breactionpotentials(SFAPs)propagatingoverthemuscle?bresbelongingtotheMU.ThesourceoftheSFAPisthe?uctuationinNaCandKCaroundthecellmembrane,whichgeneratesanintracellularactionpotentialthatthencausesatransmembranecurrent.Owingtovolumeconduction,anextracellularpotential?eldcanberecordedbyextracellularelectrodes.Asaresultofmotorneuron?ring,anactionpotentialisgeneratedatthemotorendplatewherethenerveendingsconnecttothemuscle?bre(typicallyaroundthemiddleofthe?bre).Theactionpotentialpropagateswithaspeci?cCVtowardsbothendsofthe?bre(?gure1).Themotorendplatesofallmuscle?breswithinamusclearetypicallylocatedwithinasmallbandcalledtheIZ.

TheSFAPsfromeachmuscle?breofthesameMU?realmostsimultaneously.TheshapeofoneindividualMUAPdetectedontheskinistherefore,tosomeextent,affectedbythespatialarrangementofthemotorendplates,butalsobyPhil.Trans.R.Soc.A(2009)

Downloaded from http://wendang.chazidian.com on May 31, 2010

SignalprocessingofthesEMG

nerve

MUAP distribution

over the skinfiring patternMUAP train, sEMG of 1 MUsEMG at the same location

sEMG at other locations

time (500ms)339

muscle fibres from

1 MU

Figure1.TypicalexampleofabipolarMUAPdistributionovertheskinsurfaceabovethemusclebicepsbrachiiresultingfromasuper?cialMU,eight?ringsofthesameMUresultinginanMUAPtrainrecordedbyoneelectrodepairandthesEMGresultingfromseveralactiveMUsrecordedatthesametimeandbythesameandthreeotherelectrodepairsduringanisometriccontractionatapproximately5%ofthemaximalvoluntarycontraction.TheapproximatelocationoftheMU’smuscle?bresandIZareillustrated.Firinginstances,theMUAPtrainandtheMUAPdistributionareobtainedbydecomposition,asdescribedin§7.

themuscle?brethickness,thechemicalcharacteristics(?bretype),thestateofthe?bres(fatigue),thedepthofthe?breswithinthemuscleandtheirdistancesfromthedetectionelectrode.EachMUAPcontributestotheamplitudeandfrequencycontentsofthesEMG(interferencepatternsEMG).SincelargeMUscausehigh-amplitudeMUAPs,andsincetissueactsasalow-pass?lter,largeMUsandMUsclosetotheelectrodegiveahighercontributiontotheamplitudeandfrequencycontentsfromasEMG.Amotorneuronusually?reswithacertainrateresultinginMUAPtrains(?gure1).The?ringsarenotstrictlyperiodicandthe?ringintervalscanbedescribedbyaGaussiandistribution.Atlowcontractionlevels,theaverage?ringratecanbeaslowasafewpulsespersecond(pps),andtheratetypicallyincreaseswithactivationleveluptoapproximately30pps.Withincreasingactivationlevel,thenumberofactiveMUsalsoincreases.Inslowvarying,non-fatiguingrampcontractions,suchMUrecruitmenttakesplacethroughthesizeprinciple:MUswithsmallneuronsarerecruited?rst,afterwhichMUswithgraduallylargerneuronsarerecruited.MUswithsmallneuronsgenerallycomposesmall,slow,weakandrelativelylow-fatigable?bres.Undermostlowandmoderatecontractions,thedifferentMUs?realmostindependentlyfromeachother.However,withincreasingcontractionleveland,asaresultoffatigue,trainingorsomediseases,the?ringscanbecomemoretemporallyrelated,calledMUsynchronization.

Traditionally,asEMGisrecordedasthedifferencebetweentwoelectrodes(bipolarorsingledifferentialelectrodecon?guration)placedontheskinabovetheactivemuscle,alongthe?breorientation,preferablyawayfromtheIZandtendonregion.Typically,electrodeshaveadiameterofapproximately10mmandaninter-electrodedistanceofapproximately20mm.Thedifferencebetweenthetwosignalsisampli?ed,eliminatingthecommonsignalsfromelectricaldevices(mainlypowermains)andmoredistantmuscles.ThisgivesinformationPhil.Trans.R.Soc.A(2009)

Downloaded from http://wendang.chazidian.com on May 31, 2010

340J.S.Karlssonetal.

fromalargepartofthemuscle.Applicationofmultipleelectrodes,calledmulti-channelsurfaceelectromyography,usingtwo-dimensionalhigh-densityelectrodegridsystemswithsmallelectrodes(approx.1.5mm)andsmallinter-electrodedistances(approx.5mm),allowsamoredetailedinvestigationoftheactivemuscle.Inparticular,byapplyingspatial?lterstothedetectionsystem,theactivityofdifferentregionsofamusclecanbespatiallyseparated(spatialselectivity)andindividualMUAPscanbeinvestigated.Inaddition,cross-talk,i.e.activityfromdistantbioelectricalsignals,canbereduced.Inprinciple,thespatial?lterscanbeobtainedbyusingdifferentrecordingcon?gurations.However,tobemore?exible,oftenmonopolarcon?gurations(electrodesabovethemuscleofinterestwithrespecttoacommonreferenceplacednearby,abovenon-activetissue)areusedfordatacollection,andthenspatial?lteringiscarriedoutafterdigitalization.See§3bformoreinformationonthistopic.

TheindividualMUAPstendtohaveaconsistentshapeinhealthy,non-fatiguedpersonsandcanberegardedasquasi-deterministic.However,asthenumberofindependently?ringMUsincreases,thesEMGiswelldescribedasaband-limitedGaussian-distributedstochasticprocesswithzeromean(duetothecentrallimittheorem).TheamplitudeoftherecordedsEMGistypicallyintherangeofafewmillivolts(peaktopeak),andthefrequencyspectrumrangesfromapproximately10to500Hz,withthedominantenergyinthe20–150Hzrange.ThesEMGspectrumisin?uencedbytwomajorfactors(recordingfactorsexcluded):(i)inthelow-frequencyrange(5–40Hz),theMU?ringrateand(ii)inthehigh-frequencyrange(above40Hz),theformoftheindividualMUAPs(Basmajian&DeLuca1985).Thesamplingfrequencyisoftensetat1or2kHz.Inordertoavoidaliasingofnon-sEMGsignals,accordingtotheNyquisttheorem,theanaloguesignalsarelow-pass?lteredatapproximately500Hz.Beforedigitizing,thesignalneedssignalconditioning,i.e.ampli?cationand?ltering.Formoredetailedinformation,seeMerletti&Parker(2004).

3.Pre-processingmethods

TherecordedsEMGisin?uencedbytechnicalfactors(acquisitionequipmentnoise,electrode–skincontactproperties,etc.)inadditiontophysiologicalfactors.Therefore,priortophysiologicalinterpretationorfurtherprocessing,itisnecessarytoassesswhetherornotthesignalisofsuf?cientquality.Pre-processingintheformof?lteringisusuallycarriedouttoreducethein?uenceofthetechnicalfactors.Thesetwoimportantaspectsarepresentedin§3a,b.

(a)Signalqualityassessment

Classically,sEMGqualityisestimatedbytheelectrode–skinimpedance,whichismeasuredpriortosignalacquisition.Thisgivesinformationontheconditionofthecontactbetweentheelectrodesandtheskin.However,contactandelectrode–skinimpedancemaychangeduringacquisition.Therefore,itisdesirabletobasethesignalqualityassessmentontheacquiredsEMGsignalsdirectlyandonshortperiodsoftime.

Whilethesignalqualitycouldbeassessedbyvisualinspection,thiswouldbeverylaboriousinlong-termmeasurementsofsignalsand,especially,inmulti-channelrecordedsignals.ThereareonlyafewmethodsintheliteratureforPhil.Trans.R.Soc.A(2009)