Science Arts & Métiers (SAM)
is an open access repository that collects the work of Arts et Métiers Institute of
Technology researchers and makes it freely available over the web where possible.
This is an author-deposited version published in:
https://sam.ensam.eu
Handle ID: .
http://hdl.handle.net/10985/8923
To cite this version :
Hélène PILLET, Morgan SANGEUX, Jérôme HAUSSELLE, Rami EL RACHKIDI, Wafa SKALLI
-A reference method for the evaluation of femoral head joint center location technique based on
external markers - Gait and Posture - Vol. 39, p.655-658 - 2014
Any correspondence concerning this service should be sent to the repository
A
reference
method
for
the
evaluation
of
femoral
head
joint
center
location
technique
based
on
external
markers
He´le`ne
Pillet
a,*
,
Morgan
Sangeux
b,c,d,
Je´roˆme
Hausselle
a,
Rami
El
Rachkidi
a,
Wafa
Skalli
aaArtsetMetiersParisTech,LBM,151bddel’hopital,75013Paris,France b
RoyalChildren’sHospital,Melbourne,Australia
c
UniversityofMelbourne,SchoolofEngineering,Australia
d
MurdochChildrenResearchInstitute,Melbourne,Australia
1. Introduction
Anaccuratemethodofjointcenterlocalizationisessentialfor kinematic and kineticcalculations in movementanalysis [1].To estimatethelocationofthehipjointcenter(HJC),previousstudies haveusedpredictive[2,3],functional[4,5]ormedicalimagebased
[6–9] methods. Predictive and functional methods offer easy integration into existing clinical assessment but their accuracy needs to be established before implementation. Medical image basedtechniquesmayprovideaccurate3Dpositionofjointcenters inthecoordinatesystemoftheimagingdevice.Tobeuseful,these coordinatesneedtoberegisteredtothepelviscoordinatesystem definedformovement analysisfromexternalskinmarkers. This criticalstepraisestwoissues.First,whenexternalmarkerswere notavailableonthemedicalimages,researchershavetoassumethat externalmarkerswereexactlypositionedoverbonylandmarksthat canbelocalizedonthemedicalimages.Anydiscrepancybetween externalmarkerspositionsandbonylandmarksaffectstheaccuracy of theHJCscoordinatesinthe pelviccoordinatesystem.Second, whenexternalmarkerswereavailableonthemedicalimages,the subjects’postureintheimagingdevicemaybedifferenttothestatic calibrationpostureformovementanalysis.Forexample,whenthe
subjectsarelyingsupine[6]majorbiasmaybeexpectedfromsoft tissueartifactbetweenthelyingsupineandstandinguppostures.
Inthiscontext,anewsystem,EOS1
(EOSImaging–France), basedonlow-dosebi-planarX-raystakeninastandingposition
[10]togetherwithexternalmotioncapturemarkersmayserveas areference.Imagesfromthissystemareusedtoreconstruct3D bonesmodelswhichcanberegisteredtoexternalmarkersandin theappropriatesubject’spostureforstaticcalibration.Itisthen possibletoassesstherelationshipbetweentheexternalmarkers based coordinate systems and the bones models. In clinical routine,thesystematicuseofX-rays,evenlowdose(6–9times lessthan conventionalX-rays),remains problematic.However, depending on the accuracy, it may be suitable to validate alternativemethodsofHJCslocalization.Theaimofthepresent study was to evaluate the accuracy of hip joint center (HJC) localizationusingtheEOS1
system.
2. Materialandmethod
2.1. Subjects
Seventeenvolunteers(11malesand6females)were consid-ered after informed consent and ethical committee approval. Subjectshadameanweightof75.8kg(range54.3–102kg),mean height1.74m(range1.52–1.92m)andBodyMassIndexes(BMI) rangedfrom17to33. Keywords: BiplanarX-rays EOS Gaitanalysis Biomechanics Jointcenter ABSTRACT
Accuratelocalizationofjointcentersisessentialinmovementanalysis.However,jointcenterscannotbe directlypalpatedandalternativemethodsmustbeused.Toassesstherelativemeritsofthesemethods,a medicalimagebasedreferenceshouldbeused.TheEOS1
system,anewlowdosebi-planarX-rays imagingtechniquemaybeconsidered.Theaimofthisstudywastoevaluatetheaccuracyofhipjoint center(HJC)localizationusingtheEOS1
system.Seventeenhealthyyoungadultsparticipatedinthe study.FemoralheadsandpelvicexternalmarkerswerelocalizedusingtheEOS1
systemandtheHJCs wereexpressedinthemovementanalysiscoordinatesystem.Resultsshowedthatexternalmarker localizationwasreliablewithin0.15mmfortrainedassessors.MeanaccuracyforHJClocalizationwas 2.9mm(SD:1.3,max:6.2).TheEOSbasedmethodthereforeappearedreliableandmaybeusedfor femoralheadlocalizationorasareferencetoassesstheaccuracyofothermethodsforHJClocalization.
* Correspondingauthor.Tel.:+33144246338;fax:+33144246366. E-mailaddress:helene.pillet@ensam.eu(H.Pillet).
2.2. Protocol
Volunteerswereequippedwithfour14mmreflectivemarkers placedontheanteriorandposterioriliacspinelandmarksofthe pelvis.Foreachsubject,twoacquisitionsweretakenwiththeEOS1
(Fig.1)inastandardstandingposturedescribedbyChaibietal.
[11].EachacquisitionconsistedoftwosimultaneousX-rayimages in the sagittal and coronal plane.Acquisitions were performed 5minaparttoallowthesystemtocooldown.Three-dimensional reconstructionofthefemurwasperformedasdescribedbyChaibi et al.[11].Asphere wasfittedin theleastsquare sensetothe femoral head regionof themodel tolocalize thefemoral head centerintheEOS1
coordinatesystem.Threedimensionalpositions ofthepelvicmarkerswerealsocomputed:theprojectedimagesof a14mmdiameterspheremodelweremanuallyadjustedonthe twosetsofbi-planarX-raystomatchreflectivemarkerscontours.
2.3. Repeatabilityofmarkersidentification
Datafrom4ofthesubjectswereused(1maleand1femalewith aBMI>25,1maleand1femalewithaBMI<25).Markersfrom thefirstpairofimageswereconsidered.Twotrainedoperators (bioengineers)independentlyplacedthreetimestheentiresetof markers.ThemethoddescribedbyGlu¨eretal.[12]wasusedto calculatetheinter-operatorrepeatabilityofmarkerlocalizationon theimages.
2.4. Accuracyoffemoralheadregistration(Fig.2)
The right and left femoral heads were localized for each acquisition(FH1EOS1andFH2EOS2).Wecalculatedthe
transforma-tiontomapthepelvicmarkersfromthefirstacquisitiontothe
secondacquisitionintheleastsquaresense[13].The transforma-tionwasthenappliedtothepositionofthefemoralheadfromthe firstacquisitiontomapitontothesecondacquisitionwhichgives FH1EOS2.Accuracyofthefemoralheadregistrationwascalculated
asDFH,thedistancebetweenFH1EOS2 andFH2EOS2.Itshouldbe
notedthatDFHincludesaccuracyoffemoralheadlocalizationand
repeatabilityofpelvismarkerslocalization.
3. Resultsanddiscussion
Inter-operatorvariabilityinmarkerslocalizationwas0.15mm. Uncertaintywasmainlyrelatedtovisibilityofsomeofthemarkers duetoboneormarkersuperposition,andsometimestothelackof sphericityof themarkers ontheimages.ThedistanceDFHwas
homogeneousforleftand rightHJCand2.9mminaverage(SD: 1.3).Therewasonlyoneoutlier(6.2mm,Table1,S8left)over34 measurements.
TheEOSmethodappearsmorereliablethanotherpublished methods. Ina recent studydescribing a 3D ultrasoundmethod
[14], authors reported an accuracy of 4mm for the distance betweentheleftandrightfemoralheadswithrespecttomagnetic resonanceimaging(MRI).However,sinceMRIimagesweretaken lyingsupineandexternalmotioncapturemarkerscannotbeseen ontheimages,theauthorswereunabletoassesstheaccuracyof thefemoralheadlocalizationwithrespecttothepelviccoordinate system.Leardinietal.usedasimilartechniquebasedX-ray(called RSA)tocomparethepositionoftheHJCoutputbypredictiveand functional methods [4]. The authors reported a comparable accuracy for the marker localization (0.2mm) from the RSA technique but didnot reporttheaccuracy of thefemoral head localization.However,thetechnique describedinLeardinietal. studyinvolveslevelsofionizingradiationthataretoohightobe
Fig.1.Twopairsofradiographiesforonesubject(thefirstpaircenteredontherightlowerlimbandthesecondpaircenteredontheleftlowerlimb).Pelvismarkersarevisible onbothX-rays.
used in extensive data collection and necessitated the use of tantalum ballsinstead ofexternal markers. Itis only since the EOS1
system, with low dose radiation, that such stereoradio-graphytechniquesmaybeenvisagedforextensivestudies.
Typicalaccuracyofnon-imagebasedmethodstolocatethehip joint center range from 13mm to 30mm [15]. The current technique,withanaverageand directlycomparableaccuracyof 2.9mminaverage seemsappropriatetoserveasa reference.A greatadvantageofthisapproach,i.e.simultaneousvisualizationof femoral headand externalmarkers,is thatit isnotaffectedby
subjectspecificitiessuchasBMIorunusualanatomy.Ittherefore constitutesastrongbaseeitherfordirectmeasurementofthehip jointcenterorforevaluationofalternativemethodsofestimation.
Acknowledgments
TheauthorsaregratefultoVICON(OMG-UK)fortheloanofa motioncapturesystemnecessaryfortheoverallstudy.
Conflictofintereststatement
Theauthorsdonothaveanyfinancialorpersonalrelationships with other people or organizations that could inappropriately influencethemanuscript.
References
[1]StagniR,LeardiniA,CappozzoA,BenedettiMG,CappelloA.Effectsofhipjoint centremislocationongaitanalysisresults.JBiomech2000;33:1479–87.
[2]BellAL,PedersenDR,BrandRA.Acomparisonoftheaccuracyofseveralhip centerlocationpredictionmethods.JBiomech1990;23:617–21.
[3]DavisRB,OunpuuS,TyburskiD,GageJR.Agaitanalysisdatacollectionand reductiontechnique.HumMovSci1991;10:575–87.
[4]LeardiniA,CappozzoA,CataniF,Toksvig-LarsenS,PetittoA,SforzaV,etal. Validation ofa functionalmethodfortheestimation ofhip jointcentre location.JBiomech1999;32:99–103.
[5]EhrigRM,TaylorWR,DudaGN,HellerMO.Asurveyofformalmethodsfor determiningfunctionaljointaxes.JBiomech2007;40:2150–7.
[6]OtakeY,SuzukiN,HattoriA,HagioK,SuganoN,YonenobuK,etal. Four-dimensionalmodel ofthelower extremityafter total hiparthroplasty.J Biomech2005;38:2397–405.
[7]KawakamiH,SuganoN,YonenobuK,YoshikawaH,OchiT,HattoriA,etal.Gait analysissystemforassessmentofdynamicloadingaxisoftheknee.Gait Posture2005;21:125–30.
[8]OhnishiT,SuzukiM,NawataA,NaomotoS,IwasakiT,HaneishiH. Three-dimensionalmotionstudyoffemur,tibia,andpatellaatthekneejointfrom bi-planefluoroscopyandCTimages.RadiolPhysTechnol2010;3:151–8.
Fig.2.Schematicoftheprocesstoobtainaccuracydata.
Table1
Accuracyofpelvismarkersbasedregistrationofthefemoralheadfor17subjects: distance(mm)betweenthefemoralheadregisteredfrompelvismarkersandthe femoralheadcomputedfromthefemurmodelforrightandleftsides.
Subjects Distanceforthe rightfemoralhead(mm)
Distancefortheleft femoralhead(mm) 1 2.0 3.2 2 3.7 3.6 3 3.2 1.8 4 3.3 2.7 5 0.7 1.5 6 2.8 1.7 7 2.1 1.5 8 3.0 6.2 9 1.0 1.8 10 4.5 4.4 11 0.9 2.5 12 3.8 2.8 13 4.7 4.5 14 4.2 2.1 15 4.1 3.7 16 1.7 3.0 17 3.0 3.6 Mean 2.9 3.0 Max 4.7 6.2
[9]SholukhaV,LeardiniA,SalviaP,RoozeM,SintSV.Double-stepregistrationof invivostereophotogrammetrywithbothinvitro6-DOFselectrogoniometry andCTmedicalimaging.JBiomech2006;39:2087–95.
[10]DuboussetJ,CharpakG,SkalliW,DeguiseJ,KalifaG.EOS:Anewimaging systemwithlowdoseradiationinstandingpositionforspineandbone&joint disorders.JMusculoskeletRes2010;13:1–12.
[11]ChaibiY,CressonT,AubertB,HausselleJ,NeyretP,HaugerO,etal.Fast3D reconstructionofthelowerlimbusingaparametricmodelandstatistical inferencesandclinicalmeasurementscalculationfrombiplanarX-rays. Com-putMethodsBiomechBiomedEng2012;15:457–66.
[12]Glu¨erCC,BlakeG,LuY,BluntBA,JergasM,GenantHK.Accurateassessmentof precisionerrors:howtomeasurethereproducibilityofbonedensitometry techniques.OsteoporosInt1995;5:262–70.
[13]So¨derkvistI,WedinPA.Determiningthemovementsoftheskeletonusing well-configuredmarkers.JBiomech1993;26:1473–7.
[14]PetersA,BakerR,SangeuxM.Validationof3-Dfreehandultrasoundforthe determinationofthehipjointcentre.GaitPosture2010;31:530–2.
[15]Sangeux M, Peters A, Baker R. Hip joint centre localization: evaluation onnormalsubjectsinthecontextofgaitanalysis.GaitPosture2011;34: 324–8.