Three-dimensional superimposition for patients with facial palsy: an innovative method for assessing the success of facial reanimation procedures

BritishJournalofOralandMaxillofacialSurgery56(2018)3–7

ScienceDirect

Three-dimensional

superimposition

for

patients

with

facial

palsy:

an

innovative

method

for

assessing

the

success

of

facial

reanimation

procedures

C.

Sforza

,

E.

Ulaj

,

D.M.

Gibelli

,

F.

Allevi

,

V.

Pucciarelli

,

F.

Tarabbia

,

D.

Ciprandi

,

G.

Dell’Aversana

Orabona

,

C.

Dolci

,

F.

Biglioli

a_{FunctionalAnatomyResearchCenter(FARC),LaboratoriodiAnatomiaFunzionaledell’ApparatoStomatognatico(LAFAS),LaboratoriodiAnatomia} Funzionaledell’ApparatoLocomotore(LAFAL),DipartimentodiScienzeBiomedicheperlaSalute,FacoltàdiMedicinaeChirurgia,Universitàdegli StudidiMilano,viaMangiagalli31,I-20133Milano,Italy

b_{Maxillo-FacialSurgicalUnit,OspedaleSanPaolo,DipartimentodiScienzedellaSalute,FacoltàdiMedicinaeChirurgia,UniversitàdegliStudidi} Milano,Italy

c_{DivisionofMaxillofacialSurgery,DepartmentofNeurosciences,ReproductiveandOdontostomatologicalSciences,UniversityofNaples“FedericoII”,} Naples,Italy

Accepted25November2017

Availableonline6December2017

Abstract

Facialpalsyisasevereconditionthatmaybeamelioratedbyfacialreanimation,butthereisnoconsensusabouthowtojudgeitssuccess.

Inthisstudyweaimedtotestanewmethodforassessingfacialmovementsbasedon3-dimensionalanalysisofthefacialsurfaces.Eleven

patientsagedbetween42and77yearswhohadrecentlybeenaffectedbyfacialpalsy(onsetbetween6and18months)weretreatedbyan

operationbasedontripleinnervation:themasseterictotemporofacialnervebranch,30%ofthehypoglossalfibrestothecervicofacialnerve

branch,andthecontralateralfacialnervethroughtwocross-facesuralnervegrafts.Eachpatienthadfivestereophotogrammetricscans:atrest,

smilingonthehealthyside(facialstimulus),biting(massetericstimulus),movingthetongue(hypoglossalstimulus),andcorner-of-the-mouth

smile(MonaLisa).Eachscanwassuperimposedontothefacialmodelofthe“rest”position,andthepoint-to-pointrootmeansquare(RMS)

valuewasautomaticallycalculatedonboththeparalysedandthehealthyside,togetherwithanindexofasymmetry.One-wayandtwo-way

ANOVAtests,respectively,wereappliedtoverifythesignificanceofpossibledifferencesintheRMSandasymmetryindexaccordingtothe

typeofstimulus(p=0.0329)andside(p<0.0001).RMSdifferedsignificantlyaccordingtosidebetweenthefacialstimulusandthemasseteric

oneontheparalysedside(p=0.0316).Facialstimulusevokedthemostasymmetricalmovement,whereasthemassetericproducedthemost

symmetricalexpression.Themethodcanbeusedforassessingfacialmovementsafterfacialreanimation.

©2017TheBritishAssociationofOralandMaxillofacialSurgeons.PublishedbyElsevierLtd.Allrightsreserved.

Keywords:facialpalsy;facialreanimation;stereophotogrammetry;3D-3Dsuperimposition

Introduction

Facial palsy is a severecondition that has several causes andmaybeacomplicationorexpectedoutcomeofspecific operationsonthecranialbaseandbrain.1

∗_{Correspondingauthor.Tel.:+390250315385,fax:+390250315387.}

E-mailaddress:chiarella.sforza@unimi.it(C.Sforza).

Facial paralysis greatly affects the quality of life of affectedpatients,bothclinicallyandsocially.2,3 _Atpresent,

surgicalfacialreanimationisbasedonprovidinganew neu-ralstimulus,andthehypoglossalandmassetericnervesare themostcommonlyused.4,5 Themassetericnerve,andless sothehypoglossus,partiallychangetheirfunction(smiling withoutclenchingtheteeth)–socalledcerebraladaptation6 – thoughtheiroriginal functionstill providesthe maximal

https://doi.org/10.1016/j.bjoms.2017.11.015

stimulus: smiling plus clenching the teeth or pushing with the tongue against lower incisors leads to a wider movement.

Thistype of smile is seldom spontaneous, however,as onlythefacialnervecanbeactivatedbyemotions.1 Often one or more branches of the contralateral healthy facial nervecanthereforebeusedtoevokea“qualitative” improve-ment of facial expression through the “cross-face” nerve grafts.7

Surprisinglyatpresentthereisnomethodthatisthe“gold standard” forassessingthe restored facialfunctions.2 Tra-ditional methods for verifying the improvement of facial activity are clinical classifications such as the House-Brackmannscale,8 _{whichiseasytousebutprovides only}

qualitative evaluations and is observer-dependent.9,10 An attempttointroduceaworld-wideassessmentmethodisthe e-FACEevaluationproposedbytheHarvardfacialparalysis team.11Itisbeingadoptedbymanyotherteams,andhasthe greatmeritof permittingcomparisonsbetweenunits –the draw-backisstillthelackofobjectivitybecauseitisbased onobservers’evaluations.

Withtimeandtheintroductionofmodern3-dimensional acquisition of images and elaboration systems, proce-dures in the sensitive field of facial palsy have been reported, through the 3-dimensional analysis of patients’ faces.2,12,13 However,existingstudieshavesofaranalysed the facial movements almost onlythrough the dislocation of landmarks,whichhaslimitedthe evaluationtoselected points. However, now the 3-dimensional analysis of faces allows research workers to make more detailed analyses (for example, through the recording and superimposition of 3-dimensional facial models and calculation of point-to-point distances between all the facial surfaces). This approach has already been reported in different types of research.14,15

In thisstudywe reportanewmethodfor assessingthe successoffacialreanimationsurgerythrough3-dimensional recordingandsuperimposition;theresultsmayenable clin-icians to develop new objective and quantitative methods usefulinmaxillofacialsurgicalpractice.

Patientsandmethods

Data were collected for 11subjects (mean (range) age58 (42–77) years)who hadunilateral facial palsy, inmost of casesfollowingexcisionofanacousticneurinoma(Table1). Themean(SD)timebetweenthefacialnervebeingaffected andtheoperationwas11(3)months,whilebetweenoperation and3-dimensionalanalysisitwas24months(10)months. Allpatientsweretreatedbetween2013and2016byasingle operationbasedontripleinnervation:end-to-endmasseteric totemporofacialbranchneurorrhaphy,side-to-end hypoglos-sustocervicofacialbranchneurorrhaphy,andtwocross-face suralnervegrafts(end-to-endattheproximalcoaptationand end-to-sideatthedistalone).

Before data were collected, all patients were given a detaileddescriptionoftheprocedureandsignedaninformed consentformthathadpreviouslybeenapprovedbytheethics committee of the University of Milan Medical School in accordance withthe standards of the 1964 Declaration of Helsinki.Noprocedurewasinvasive,dangerousorpainful, andinvolvedminimaldiscomfort.

A series of 50 reference points were marked on each face according to a set of landmarks already described elsewhere.16,17Eachpatientsfacewasscannedfivetimesby stereophotogrammetry(VECTRA-3D®:CanfieldScientific, Inc.,Fairfield,NJ).Thefirstscanwastakeninthe“rest” posi-tion,andthenscansweretakenduringaposed“smile”that wasevokedbythethreefunctionalmanoeuvresrecognised asstimuliforthecorrespondingnervousconnections (smil-ingonthehealthysideforthecross-faceprocedure,bitingfor massetericneurorrhaphy,andpushingwiththetongueagainst the lower incisors for hypoglossal neurorrhaphy). Finally, theywererequestedtoproducethemostnatural corner-of-the-mouthsmile(MonaLisa)usingallthestrategiesthatthey hadlearned.

The 3-dimensional reconstructions of the smiling face obtainedthroughthefourdifferentsmilingmanoeuvreswere recorded on the corresponding neutral one for a total of foursuperimpositionsforeachsubject.Toconstructaproper superimposition,afacialareaofinterestwassegmentedin

Table1

Clinicaldataofthe11patientsselectedforthestudy.ThepreoperativeHouse–Brackmannscorewassixforallpatients.

Sex Age

(years)

Diagnosis Timebetweenlesionand

operation(months)

Timebetweenoperation

and3Danalysis(months)

Postoperative

House–Brackmannscore

Female 60 Acousticneurinoma 14 15 2

Male 49 Acousticneurinoma 10 14 3

Male 69 Acousticneurinoma 13 43 3

Female 52 Acousticneurinoma 13 29 2

Male 59 Acousticneurinoma 6 22 2

Male 77 Acousticneurinoma 8 23 2

Female 53 Caraccident 18 13 3

Female 43 Acousticneurinoma 10 19 2

Female 68 Acousticneurinoma 10 14 2

Female 42 Acousticneurinoma 13 32 2

Female 68 Neoformationofthepetrous

portionoftemporalbone

Fig.1.Stepsof3-dimensionalsuperimposition:A)3-dimensionalfacialmodelofthepatientatrest.B)3-dimensionalmodelofthepatient(withMonaLisa

or“corner-of-the-mouth”smile).C)Superimpositionaccordingtotheleastpoint-to-pointdistancebetweenthetwomodels.D)Chromaticmapofdistances

betweenthetwomodels:greenareasareunchanged,blueareasaremoreprominentinthesmilingmodelthanintherestposition,andviceversaforthered

andyellowareas.Inthiscasetherightsideistheparalysedoneandshowsmostgreencolouration.

eachmodel.18Theareaofinterestwasautomatically super-imposedbythesoftwaretoreachtheleastdistancebetween pointsoftheentiresurfaces(Fig.1).

Afterthe recordingbetween the two surfaceshadbeen reached,thefacialmodelswerefurthersegmentedtodivide therightfromtheleftsideaccordingtosevenmidline land-marks: (trichion, sellion, pronasale, subnasale, sublabiale, pogonion,menton).

WethenusedtheMirror®Vectrasoftware(Canfield Sci-entific,Inc.,Fairfield,NJ)tocalculatethepoint-to-pointroot mean square (RMS)value between the neutral expression modelandthedifferenttypesofsmileontheparalysedand healthysides,separately.Inaddition,weextractedan asym-metryindexfromtheRMSvalues,astheabsolutevalueof thefollowingformula:(RMShealthyside−RMSparalysed side)*100/RMShealthyside.

Theentireprocedurefrom segmentationof theareasof interesttothecalculationoftheRMSvalueswasrepeated for24superimpositionsbythesameoperatorandbyanother observer to verify intraoperator and interoperator error, respectively,usingtheBland–Altmantest.

TheJarque–BeratestandBartletttestwereused to ver-ify the normal distribution andhomoscedasticity of RMS valuesandoftheasymmetryindex,respectively.Bothtests weredoneusingMatlab®software.WhenRMSdatadidnot passtheJarque–BeraorBartletttest, thenaturallogarithm of the RMSvalue andasymmetry index was used for the analyses, after having verified the normal distributionand homoscedasticityofthetransformeddata.

Resultswerethenanalysedusingatwo-wayANOVAto verifysignificantdifferencesbetweenRMSvaluesaccording tothe side, typeof stimulus, andtheir interaction. Proba-bilitiesof less than 0.05 wereaccepted as significant. For bothtests, post-hoccomparisonsweremadeseparatelyfor theparalysedandhealthysideusingaone-wayANOVA.

Symmetry indices were analysed through one-way ANOVAtoassessthesignificanceofdifferencesaccording tothetypeofstimulus,andpost-hoctestsweredonewhen appropriate.

Results

The extraction of RMS values showed intraobserver and interobserverrepeatabilityof97%.

ResultsareshowninTable2.Onthehealthysideofthe face, cross-facestimulus(smilingonthe healthyside)and massetericstimulus(biting)reachedthehighestRMSvalues. Thesamevariableswerelowerontheparalysedsidethanon thehealthyone:themassetericstimulusreachedthehighest RMSdistancerelativetotherestposition,followedbythe hypoglossalone.TheMonaLisasmilereachedintermediate scoresforboththehealthyandparalysedsides.

RMSvaluesintheeightgroupsfollowedanormal distri-butionbutdidnothavesimilarvariances;ontheotherside,the naturallogarithmofRMSvaluesshowedthattheywere nor-mallydistributedandhadsimilarvariances,soweappliedthe statisticalanalysesusingtheANOVAtothenaturallogarithm ofRMSvalue(logeRMS).

Significantdifferenceswerefoundaccordingtoside(F: 20.91;p=<0.0001;df:1;80)andtypeofstimulus(F:3.06; p=0.0329;df:3;80).Sidexstimulusinteractiondidnotreach

significance (F: 2.2; p=0.0945; df: 3;80). On the healthy side,post-hoctestingfailedtoverifysignificantdifferences accordingtotypeofstimulus(F:2.2;p=0.1030;df:3;40). Ontheotherside,significantdifferenceswerefoundbetween thefacialandmassetericstimulusontheparalysedside(F: 3.25;p=0.0316;df:3;40).

The highest asymmetry index was shownby the facial cross-facestimulus,whereasthemassetericstimulusproved toevokethemostsymmetricalmovementsoftheface.The asymmetryindexwasnormallydistributedandhomogenous according to variance. Significant differences were found according to the type of stimulus(F: 3.64; P: 0.0237; df: 3;30), although on posthoctesting therewere differences onlybetweenthefacialandmassetericstimuli.

Discussion

Facialpalsyhasobviousphysicalconsequencesintheform oforalincompetenceandcorneallesions,andtheasymmetry anddistortionoffaceoftenleadtosocialisolation.1,3,4Facial reanimationhasprovedtobeanoptiontominimisetheeffects ofthefacialpalsy,butwestilldonothaveauniversallyagreed quantitativemethodfortheassessmentoffacialmovements. Existingclinicalscalesarequalitative,subjective,anddonot quantifythefacialmodificationsevokedbydifferentstimuli, althoughtheycanpromotecomparisonsbetweencentres (e-FACE).12

The introduction of modern 3-dimensional acquisition and elaboration systems has enabled research workers to improveanalysesoffacialmovement,asshownbyPopatetal whousedastereophotogrammetricmotionanalyser.19,20The mainlimitationofthismethodisthatitrestrictsthekindof movementsthatcanbeanalysed.19,20Asimilarmethodwas alsousedbyOkadathroughlaserscanning.21

We have developed a method for the quantitative assessment of facial mimicry based on the 3-dimensional displacement of landmarks detected byan instrument that captures motion,2,22 and the protocol has proved to be repeatable.23 Unfortunately, the instrument is not widely used,anditisunlikelythatitsusewillbecomewidespread.24 Indeed,wehavealsorecentlydeviseda2-dimensional photo-graphicmethodtoquantifythesuccessoffacialrehabilitation procedures,25andthispaperisafurtherstepinthatdirection. To ourknowledgethisisthefirststudythathasapplied the modern procedures of 3-dimensional facial recording

Table2

Rootmeansquarevaluesforthehealthyandparalysedsides,andasymmetryindexforallthefourelicitedstimuli.Valuesaremean(SD).

Facialcross-face stimulus Hypoglossal stimulus Masseteric stimulus Corner-of-the-mouth smile Healthyside(mm) 1.38(0.63) 0.82(0.44) 1.31(0.46) 1.13(0.53) Paralysedside(mm) 0.54(0.24) 0.67(0.31) 0.95(0.49) 0.59(0.14) Asymmetryindex(%) 62(28) 42(25) 30(22) 41(20)

and superimposition to patients affected by facial palsy: theproposedprotocoliseasilyrepeatable,contactless, and non-invasive with clear advantages in comparison with electromyography, which is usually used to assess facial movements.

Interestingly, our results confirm the common opin-ion about the three different reported surgical techniques: cross-face intervention is unable to produce important facial movements, being a “qualitative” nervous connec-tion,whereasthemassetericoneprovidesthe widestfacial modifications.1

The highest facial asymmetry is reached by the facial cross-facestimulus,asitevokesthemovementofthehealthy side,whereastheparalysedoneisminimallystimulated.On theotherside,themassetericonecreatesthemost symmetri-calexpression. TheMonaLisasmile reached intermediate levels of bothfacial movementand asymmetry. After the operation,patientssoonlearnthatthemost“natural”smile canbereachedthroughalimitedactivationofmusclesonthe healthyside(reducingthepullingeffectonthepareticside), whichincreasesthesymmetryoftheexpression.1,2

Inconclusion,thisstudydescribesanew,highly repeat-able method for assessing facial movements in patients treatedbyfacialreanimation.Itmaybeusefulnotonlyfor assessingthesuccessofsurgicaltreatment,butalsoasatest fortheprogressivetrainingofpatientsduringthefollow-up phase.

Ethicsstatement/confirmationofpatients’permission

Beforedata werecollected,we gaveadetaileddescription of the procedure to all patients, who signed an informed consentformpreviouslyapprovedbytheethicscommittee oftheUniversityofMilaninaccordancewiththestandards oftheDeclarationof Helsinki.Noprocedurewasinvasive, dangerous,orpainful,andinvolvedminimaldiscomfort.

Conflictofinterest

Wehavenoconflictsofinterest.

References

1.BiglioliF.Facialreanimations:partI—recentparalyses.BrJOral Max-illofacSurg2015;53:901–6.

2.SforzaC,TarabbiaF,MapelliA,etal.Facialreanimationwith masse-terictofacialnervetransfer:athree-dimensionallongitudinalquantitative evaluation.JPlastReconstrAesthetSurg2014;67:1378–86.

3.WalkerDT,HallamMJ,MhurchadhaS,etal.Thepsychosocialimpact offacialpalsy:ourexperienceinonehundredandtwentysixpatients. ClinOtolaryngol2012;37:474–7.

4.YetiserS,KarapinarU.Hypoglossal-facialnerveanastomosis:a meta-analyticstudy.AnnOtolRhinolLaryngol2007;116:542–9.

5.Biglioli F, Frigerio A, Colombo V, et al. Masseteric-facial nerve anastomosis for early facial reanimation. J Craniomaxillofac Surg 2012;40:149–55.

6.ManktelowRT,TomatLR,ZukerRM,etal.Smilereconstructioninadults withfreemuscletransferinnervatedbythemassetermotornerve: effec-tivenessandcerebraladaptation.PlastReconstrSurg2006;118:885–99.

7.SmithJW.Anewtechniqueoffacialreanimation.In:HuestonJH, edi-tor.TransactionsofthefifthInternationalCongressofPlasticSurgery. Sydney:Butterworths;1971.

8.MartinsRS,SocolovskyM,SiqueiraMG,etal.Hemihypoglossal-facial neurorrhaphy aftermastoiddissection ofthe facial nerve:results in 24patientsandcomparisonwiththeclassictechnique. Neurosurgery 2008;63:310–7.

9.HontanillaB,MarrèD.Comparisonofhemihypoglossalnerveversus massetericnervetranspositionsintherehabilitationofshorttermfacial paralysisusingtheFacialClimaevaluatingsystem.PlastReconstrSurg 2012;130:662e–72e.

10.HontanillaB,MarrèD,CabelloA.Facialreanimationwithgracilis mus-cletransferneurotizedtocross-facialnervegraftversusmassetericnerve: acomparativestudyusingtheFacialClimaevaluatingsystem.Plast ReconstrSurg2013;131:1241–52.

11.Banks CA, Bhama PK, Park J, et al. Clinician-graded elec-tronicfacial paralysisassessment: TheeFACE. Plast ReconstrSurg 2015;136:223e–30e.

12.RolandJrJT,LinK,KlausnerLM,etal.Directfacial-to-hypoglossal neurorrhaphywithparotidrelease.SkullBase2006;16:101–8.

13.SforzaC,GuzzoM,MapelliA,etal.Facialmimicryafter conserva-tiveparotidectomy:athree-dimensionaloptoelectronicstudy.IntJOral MaxillofacSurg2012;41:986–93.

14.GibelliD,DeAngelisD,PoppaP,etal.Anassessmentofhowfacial mimicrycanchangefacialmorphology:implicationsforidentification. JForensicSci2017;62:405–10.

15.DjordjevicJ,JadallahM,ZhurovAI,etal.Three-dimensionalanalysisof facialshapeandsymmetryintwinsusinglasersurfacescanning.Orthod CraniofacRes2013;16:146–60.

16.FerrarioVF,SforzaC,SerraoG,etal.Growthandagingoffacialsoft tissues:Acomputerizedthree-dimensionalmeshdiagramanalysis.Clin Anat2003;16:420–33.

17.FerrarioVF,SforzaC.Anatomyofemotion:a3Dstudyoffacialmimicry. EurJHistochem2007;51(suppl1):45–52.

18.CodariM,PucciarelliV,StangoniF,etal.Facialthirds—basedevaluation offacialasymmetryusingstereophotogrammetricdevices:Application tofacialpalsysubjects.JCraniomaxillofacialSurg2017;45:76–81.

19.Popat H, Richmond S, Playle R, et al. Three-dimensional motion analysis—anexploratorystudy.PartI:assessmentoffacialmovement. OrthodCraniofacRes2008;11:216–23.

20.Popat H, Richmond S, Playle R, et al. Three-dimensional motion analysis—anexploratorystudy.PartII:reproducibilityoffacial move-ment.OrthodCraniofacRes2008;11:224–8.

21.Okada E. Three-dimensional facial similations and measurements: changesinfacialcontourandunitsassociatedwithfacialexpression. JCraniofacSurg2001;12:167–74.

22.SidequerskyFV,MapelliA,AnnoniI,etal.Three-dimensionalmotion analysisoffacialmovementduringverbalandnonverbalexpressionsin healthysubjects.ClinAnat2016;29:991–7.

23.SforzaC,FrigerioA,MapelliA,etal.Facialmovementbeforeandafter masseteric-facialnervesanastomosis:athree-dimensionaloptoelectronic pilotstudy.JCraniomaxillofacSurg2012;40:473–9.

24.SidequerskyFV,VerzéL,MapelliA,etal.Quantificationoffacial move-mentsbyopticalinstruments:surfacelaserscanningandoptoelectronic three-dimensionalmotionanalyzer.JCraniofacSurg2014;25:e65–70.

25.ZagoM,TarabbiaF,BassettiL,etal.Photographictechniqueforthe quantitativeassessmentoflagophthalmosandeyelidpositioninpatients withfacialpalsy.BrJOralMaxillofacSurg2017;55:548–50.