Posted at the Institutional Resources for Unique Collection and Academic Archives at Tokyo Dental College, Available from http://ir.tdc.ac.jp/
Title
Accuracy of Le Fort I osteotomy with combined computer‑aided design/computer‑aided manufacturing technology and mixed reality
Author(s) Alternative
Koyachi, M; Sugahara, K; Odaka, K; Matsunaga, S;
Abe, S; Sugimoto, M; Katakura, A
Journal International journal of oral and maxillofacial surgery, 50(6): 782‑790
URL http://hdl.handle.net/10130/5458
Right
© 2020 The Author(s). Published by Elsevier Ltd on behalf of
International Association of Oral and Maxillofacial Surgeons. This is
an open access article under the CC BY‑NC‑ND license
(http://creativecommons.org/licenses/by‑nc‑
nd/4.0/).
Description
Orthognathic Surgery
Accuracy of Le Fort I osteotomy with combined computer-aided design/computer-aided
manufacturing technology and mixed reality
M.Koyachi,K.Sugahara,K.Odaka,S.Matsunaga,S.Abe,M.Sugimoto,A.Katakura:
AccuracyofLeFortIosteotomywithcombinedcomputer-aideddesign/computer- aidedmanufacturingtechnologyandmixedreality.Int.J.OralMaxillofac.Surg.
2021;50:782–790.ã2020TheAuthor(s).PublishedbyElsevierLtdonbehalfof InternationalAssociationofOralandMaxillofacialSurgeons.Thisisanopenaccess articleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc- nd/4.0/).
M.Koyachi1,K.Sugahara1,2, K.Odaka3,S.Matsunaga2,4, S.Abe2,4,M.Sugimoto1,5, A.Katakura1,2
1DepartmentofOralPathobiologicalScience andSurgery,TokyoDentalCollege,Tokyo, Japan;2OralHealthScienceCentre,Tokyo DentalCollege,Tokyo,Japan;3Departmentof OralandMaxillofacialRadiology,Tokyo DentalCollege,Tokyo,Japan;4Departmentof Anatomy,TokyoDentalCollege,Tokyo, Japan;5OkinagaResearchInstitute InnovationLaboratory,TeikyoUniversity, Tokyo,Japan
Abstract. Theaimofthisstudywastoverifythereproducibilityandaccuracyof preoperativeplanninginmaxillarepositioningsurgeryperformedwiththeuseof computer-aideddesign/manufacturingtechnologiesandmixedrealitysurgical navigation,usingnewregistrationmarkersandtheHoloLensheadset.Eighteen patientswithameanageof26.0yearswereincluded.Postoperativeevaluationswere conductedbycomparingthepreoperativevirtualoperationthree-dimensionalimage (Tv)withthe1-monthpostoperativecomputedtomographyimage(T1).Thethree- dimensional surface analysis errors rangedfrom 79.9% to 97.1%, withan averageerror of90.3%.Inthepoint-basedanalysis,theerrorsateachpointontheXYZaxeswere calculatedforTvandT1inallcases.Themediansignedvaluedeviationofall calculatedpointsontheXYZaxeswas 0.03mm(range 2.93mmto3.93mm).The medianabsolutevaluedeviationofallcalculatedpointsontheXYZaxeswas0.38mm (range0mmto3.93mm).Therewerenostatisticallysignificantdifferencesbetween anyofthepointsonanyoftheaxes.Thesevaluesindicatethatthemethodusedwas abletoreproducethemaxillapositionwithhighaccuracy.
Keywords:augmentedreality;computer-aided design; computer-assisted surgery; orthog- nathicsurgery;LeFortosteotomy.
Acceptedforpublication
Availableonline3November2020
TheLeFortIosteotomyisahighlyeffective treatment for skeletaljawdeformities. The doublesplintmodificationoftheLeFortI osteotomy,introducedbyLindorfandStein-
ha¨user1, ishighly effectiveforrecreatinga mould-based model during surgery and is used widely on a global scale. However, Mazzoni et al.2 and Sharifi et al.3 have
reportedthattheaccuracyofmaxillareposi- tioningduringtheLeFortIosteotomypro- cedure using the double splint method is dependentonthetechnicalexpertiseofthe 0901-5027/060782+09 ã2020TheAuthor(s).PublishedbyElsevierLtdonbehalfofInternationalAssociationofOralandMaxillofacialSurgeons.Thisisan openaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).
surgeon,andthaterrorsfromthepreoperative modelplanningstagecanoccurduringface- bowtransferandsplintfabrication.Tominaga etal.4reportedthattheaccuracyofmaxillo- facialsurgerycanbeimprovedbyfabricating a preoperative bone fragment positioning guide.However,evenwiththeseadditional techniques,theexpertiseandexperienceof thesurgeonplayalargeroleinthepatient prognosis.Forthisreason,thereisgreatde- mandforthedevelopmentofasystemthat accuratelyrecreatesthequantitativelyevalu- atedtreatmentplanduringsurgery.
Inrecent years, surgical disciplines have beeninthemidstofaparadigmshiftdueto therapiddevelopmentofsurgicalnavigation techniquesthatutilizedigitalequipment.Wong et al.5 conducted surgical navigation using three-dimensional(3D)mouldscreatedbya 3D printer, and this method has shown a certain degreeofsuccessinorthopaedicdisciplines.
Regarding splint-guided osteotomies, Park et al.reported goodresultsfororthognathic surgeryusingcomputer-aideddesign/comput- er-aidedmanufacturing(CAD/CAM)guides, buttheycouldbemoreaccurate6.
Reports have indicated that head- mounteddisplays (HMDs)withinstalled mixed reality(MR) techniques can con- firmthelocationofvascularpedunculated flaps,thusminimizingtheriskofhaemor- rhage from small veins7. Despite these advancements, the success or failure of surgery inthemaxillofacialareamaybe determined by small-scale differences;
therefore,aspecializedandextremelyac- curatesystemisnecessary.
Therefore,theaimofthisstudywasto investigate theaccuracyoftheLeFortI osteotomy procedure performed using CAD/CAMsurgicalguidesandnewreg- istrationmarkersforMRsurgicalnaviga- tionandtoverifythereproducibilityofthe preoperativeplan.
Materialsandmethods
Theprocedureperformedinthisstudywas conductedusingsixsteps,whichareout- lined below and shown in Fig. 1: (1) acquisitionoftheskulldatafromcomput- ed tomography (CT) and the dentition
from laser scanning; (2) treatmentplan- ning via the creation of a skull–dental composite3Dmodelandadigitaldental model, inaddition tovirtualsurgery;(3) fabricationof3Ddevicesandcreationof the Microsoft HoloLensapplication; (4) performanceoftheactualoperationbased onthevirtualoperation;and(5)evaluation oftheaccuracyoftheoperationbycom- parison ofthevirtual operation with the postoperativeCT.
Studypopulation
Detailsofthestudysubjectsaregivenin Table1.Eighteenpatientswereincluded inthisstudy;theyrangedinagefrom17to 45years,withameanageof26.0years.
ThesurgerieswereconductedintheOral SurgeryDepartmentofSuidobashiHospi- tal, Tokyo Dental Collegefrom October 2018toOctober2019.Patientswithacleft palate or craniofacial abnormality were excludedfromthisstudy. Thestudywas approved by the Ethics Committee of TokyoDentalCollege(No.794844).
LeFortIosteotomybyCAD/CAMandmixedreality 783
Fig.1. Workflowofthisstudy.
Table1. Clinicalinformationforthestudypatientsandtheresultsofthethree-dimensionalsurfaceanalysis.
Patient Age(years) Diagnosis Surgicaltreatmentformaxilla Error<2mm(%)a
1b 24 FA,CB Advancement,rightup,leftdown 97.1
2 30 RU,PL,ACB Advancement,anticlockwiserotation 89.5
3b 38 FA,CB Rightdown,leftup 92.9
4 24 RU,PL Advancement 79.9
5b 21 FA,CB Leftshift,leftdown 95.0
6 25 RU,PL,ACB Advancement 93.3
7b,c 26 FA,AOB Rightup,leftup 92.0
8b,c 36 FA,RL,DB Rightup,leftup 95.6
9b,c 28 FA,PL,AOB Rightshift,rightup 90.3
10 28 PU,RL,DB Setback,anticlockwiserotation 88.9
11b,c 45 FA,AOB Rightup 96.1
12b,c 24 FA,PL,ACB Rightdown,leftup,impaction 84.0
13c 17 RU,PL,ACB Advancement,leftshift,impaction 81.5
14c 17 PU,RL,AOB Anticlockwiserotation 82.5
15b 19 FA,PL,ACB Setback,rightshift,rightup 91.3
16 20 RU,PL,ACB Advancement 93.6
17b 26 FA,PL,ACB Rightshift,leftdown 91.1
18 20 RU,PL Advancement,rightdown,leftdown 90.0
ACB, anteriorcrossbite;AOB,anterioropenbite;CB,crossbite;DB,deepbite;FA, facialasymmetry; PL,prognathiaoflowerjaw;PU, prognathiaofupperjaw;RL,retrognathiaoflowerjaw;RU,retrognathiaofupperjaw.
aTheaveragepercentageofboneerrorsof<2mmbetweentheTvandT1imageswas90.3%.
bCaseswithtiltimprovement.
cCaseswithverticalrepositioningoftheposteriornasalspine(PNS).
Imageacquisition
CT scans (SOMATOM Definition AS;
Siemens, Forchheim, Germany) were obtained a month prior to surgery, and theCTimageswereformattedasDICOM data.Sinceaccuratereconstructionofthe dentition is difficult due to orthodontic bracketsandprostheticartefacts,adigital impressionofaplastercastwasscanned (D2000; 3Shape, Szczecin, Poland) and convertedintoSTLdataformat.TheCT DICOManddigitalimpressiondatawere transferred to Mimics (Materialise, Leu- ven, Belgium) and 3-matic (Materialise, Leuven,Belgium)andsetuptoallowthe virtualoperation.
Preoperativeprocedure—3Ddevices The CT data and dentition mould were alignedusingMimicsand3-matic.Thesu- perimpositionoftheCTdataandthedenti- tion modeldata wasperformedusingany threepointsonthetoothincisalandocclusal surfacesthatcouldbeclearlyconfirmed.The extentofthemaxillarepositioningwasde- terminedfromthesedatafollowingdiscus- sionswiththeorthodontists,andthesurgeon determinedtheosteotomyincisionlinesand designedtheosteotomydevice,reposition- ingdevice,splint,andregistrationmarkers.
ThesedatawereconvertedintoSTLformat, andeach3Ddevicewasfabricatedusinga 3Dprinter(Objet260Connex;StratasysLtd, Eden Prairie, MN, USA). The osteotomy
deviceandrepositioningdeviceweremade so that they could be attached to the splint and switchedinoroutatthisjunctionsite.No screwsoradhesiveswereusedatthejunction site; instead a ‘Koshikakearitsugi’ (i.e., a stepped dovetail splice) configuration, which isusedinshrinecarpentrytechniques,was applied.Itwasensuredthattheattachment wassecurebymaintaininganopenwindow totheuppersectionadjacenttotheanterior nasalspine(ANS)(Figs2and3).
Preoperativeprocedure—Microsoft HoloLens
Thevirtual reality(VR) applicationwas created using Unity 2017.4.17 (Unity Fig.3. 3Ddevicesmadebycomputer-aidedmanufacturing:(A)surgicalsplint;(B)osteotomydevice;(C)repositioningdevice;(D)surgicalsplint connectedtotheosteotomydevice;(E)surgicalsplintconnectedtotherepositioningdevice.Theosteotomydeviceandrepositioningguidewere customizedineachcase.
Fig.2. Computer-aideddesignofthe3Ddevices:(A)splint(yellow)andosteotomydevice(red);(B)repositioningdevice(blue);(C)registration marker.Theosteotomydeviceandrepositioningdeviceweremadesothattheycouldbeattachedtothesplintandswitchedinoroutattheupper junctionsite.Theregistrationmarkercouldbeattachedtothelowerjunctionsite.
Technologies,SanFrancisco,CA,USA), Visual Studio Community 2017/version 15.9.4(Microsoft Corp.,Redmond, WA, USA), and OpenCV 3.1 (Open Source Computer Vision Library)+ArUcosoft- ware (AVA Group, University of Co´r- doba, Co´rdoba, Spain). The hologram reconstructs theskull,the pre-and post- repositioned maxilla/mandible, the mar- kers,andthearteries/veinsfromthecon- trast CT scans. Each VR could modify movement, rotation, and transparency withgesturalmovements,whichwasuse- fulforconfirmationwhenthepatientand theapplicationwerealigned. Additional- ly, the displays of each VR could be toggled on or off. The maxillary bone could be chosen from two VR types (pre- and post-repositioning). Further- more,theoperatingfieldandtheapplica- tion could be automatically aligned by fabricatingthemarkersusedforalignment with the 3D printer, and the HoloLens recognized the markers attached to the splint in our application. The surgeon couldconfirmthemarker,andthesystem was designed withtwosurfaces (surface
size=1010cm) thatdid notinterfere withtheoperatingfield(Fig.4).
Surgicalprocedure
All surgeries were performedby a single specialist oral and maxillofacial surgeon.
ThedirectionsofthemovementsoftheLe FortIsegmentarereportedinTable1.Inall LeFortIosteotomies,anincisionwasap- pliedfromtherightmaxillarysecondpre- molartotheleftmaxillarysecondpremolar.
Thesplintwasfittedontothemaxillaandthe osteotomydevicewasmountedatthejunc- tionsite.Inaddition,ascrewwasinserted andtemporarilyfixedintothescrewhole.
Theosteotomydevicewasremovedandthe maxillawasdown-fracturedafteranosteot- omyhadbeenperformedonboththelateral andmedialsidesofthemaxilla.Theosteot- omydevicewasthenswitchedtotherepo- sitioning device, and the maxilla was repositioned so that the screw could be inserted into the screw hole, after which themaxillawasfixedusingtwometalsix- hole plates from the bilateral zygomatic buttress.Theregistrationmarkerwascon-
nectedtothesplintjunctionsite,andoncethe repositioningofthemaxillawasconfirmed in 3D to be that specified by the virtual operationwiththeHoloLens,thereposition- ingdevicewasremovedandthemaxillawas fixedtothebuttressofthepiriformmargin usingtworesorbablefive-holeplates.
Evaluation
Postoperativeevaluationswereconducted bycomparingthepreoperativevirtualop- eration3Dimage(Tv)withthe 1-month postoperativeCTimage(T1)usingMate- rialise Mimics and 3-matic. Overlaying and evaluationofthe Tv andT1images were conducted using GOM Inspect (GOM, Braunschweig, Germany). The skull, which was unaffected during sur- gery, was used asa referencefor image overlaying,andthreearbitrarypointswere automatically selected following setup8. Thereferenceplanewasestablishedusing the methods ofBadiali et al.9 and Park et al.10:the3D-CTcentrewas setasthe middlepointbetweentheporion(Po)on bothsides,theX-axiswassetasthemiddle LeFortIosteotomybyCAD/CAMandmixedreality 785
Fig.4. Intraoperativeview:(A)operators wearingHoloLensheadsetscansharethehologramand manipulateitusinggesturesand voice commands;(B) theoperating fieldandtheapplicationcanbe alignedautomaticallybyfabricating themarkers,and theHoloLensdevice recognizesthemarkers.
Fig.5. Referenceplaneandpostoperativeevaluations:(A)thereferenceplanewasestablished,wherethe3D-CTcentrewassetasthemiddle pointbetweentheporionsonbothsides,theX-axiswassetasthemiddlepointthatpassesbetweentheorbitalesonbothsides,theY-axiswassetas theverticalcranialdirectionfromtheFrankforthorizontalplane,andtheZ-axiswassetastheright-handdirectionfromthecentre;(B)a3D surfaceanalysisandapoint-basedanalysiswereusedasthemethodsofevaluation.Thepointsevaluatedwereasfollows:A-point,anteriornasal spine(ANS),posteriornasalspine(PNS),centreofthemaxillarycentralincisorrootapex,thecaninerootapexonbothsides(#13and#23),and thefirstmolarrootapexonbothsides(#16and#26).
pointthatpassesbetweentheorbitale(Or) on both sides, theY-axis was set as the verticalcranialdirectionfromtheFrank- forthorizontal(FH)plane,andtheZ-axis was set astheright-hand direction from thecentre,withtheX/Zplanesetuptobe alignedwiththeFHplane.Sinceonlythe bone surfaces werebeing compared, the metallicplatesectionsinthepostoperative CTwereremovedusingthemethodsout- lined by Badiali et al.9. A 3D surface analysisand apoint-basedanalysiswere usedasthemethodsofevaluation(Fig.5).
The absolute values and signed values wereused11.Theabsolutevaluesexpress theabsoluteaccuracyofthemethod,and the signed values express an under (+
value)orover( value)correction.
3Dsurfaceanalysis
Errors between the Tv and T1 bone surfaces were measured. Based on previous reports, the threshold value
was setat <2mm, andthe percentage
of errors under this value was calculated12–14.
Point-basedanalysis
Thepointsevaluatedwereasfollows:A- point, ANS,posteriornasalspine (PNS), centreofthemaxillarycentralincisorroot apex,thecanine rootapexonbothsides (#13 and #23), and the first molar root apexonbothsides(#16and#26).Tvand T1werecompared,andthedistanceerrors oneachaxis(Xv–X1,Yv–Y1, Zv–Z1) were measured.Themeasurements were performedtwicebythesameoperator,and theaveragevaluewasusedastheevalua- tionvalue.Theintra-classcorrelationco- efficient wasdeterminedforthetwosets ofTvmeasurementsandforthetwosetsof T1 measurements; the correlationcoeffi- cientineach casewas greaterthan0.99.
Regarding theinfluenceofpostoperative correction,therootapexwasused asthe evaluationcriterion15.
Statisticalanalysis
All statistical analyses were performed using IBM SPSS Statistics version 23.0 (IBM Corp., Armonk, NY, USA). The Mann–Whitney U-test and Kruskal–
Wallis testwereusedtocomparetheTv andT1images.Statisticalsignificancewas set atP<0.05.
Results
3Dsurfaceanalysis
Theresultsofthe3Dsurfaceanalysisare reported in Table 1. The percentage of bonemeasurementerrorwithin2mmbe- tweentheTvandT1imagesrangedfrom 79.9%to97.1%,withanaverageerrorof 90.3%. Theaccuracywas>90.0%in12 cases.
Point-basedanalysis Signedvalues
TheerrorsateachpointontheXYZaxes were calculated for Tv and T1 in all cases. Themediansigned valuedevia- tionofallcalculatedpointsontheXYZ axes was 0.03mm (range 2.93mm to 3.93mm)(Fig. 6).
Thesignedvaluewasamedianof 0.14 mm(range 1.52mmto3.93mm)onthe X-axis, 0.10mm (range 2.93mm to 3.24mm) on the Y-axis, and 0.09mm (range 2.59mmto1.42mm)onthe Z- axis.Themedianerrorforeachaxiswhen comparingTvandT1waswithin2mm.The signed value deviation for ANS varied widelyontheX-axis,from 0.42to3.93 mm.The signedvaluedeviationforPNS hadthelargestrangeontheY-axis,from 2.93to3.24mm(Fig.7).Therewereno
Absolutevalues
Themedianabsolutevaluedeviationofall calculatedpointsontheXYZaxeswas0.38 mm(range0mmto3.93mm)(Fig.8).
The absolute value was a median of 0.56mm (range 0–3.93mm) on the X-axis, 0.29mm (range 0.01–3.24mm) on the Y-axis, and 0.33mm (range 0–2.59mm) on the Z-axis (Fig. 8). The absolute value also showed that the median error for each axis (comparison between Tv and T1) was within 2mm.
Similartothesignedvalues,theaccuracy tendedtobelowerforANSontheX-axis andPNSontheY-axis(Fig.9).Compar- isonsbetweentheaxesshowednostatisti- callysignificantdifferencesbetweentheY and Z axes (P>0.05), but statistically significantdifferenceswereobservedbe- tween the X and Yaxes (P<0.05) and betweentheXandZaxes(P<0.05).
Registrationtoparticipateinthisstudy didnotresultinanydelayinsurgery,and thedurationofthesurgicalprocedurewas relativelyunaffectedbecausethesurgeon beganthesurgerywhilewearingtheHolo- Lens headset. Additionally, the surgeon couldsmoothlyoperatethesurgicalguide inallcases.Noabnormalhaemorrhaging or fractures were observed during the procedureforanyofthepatients.
Discussion
Withthedevelopmentofcomputer-aided surgicalsimulationintheoralandmaxil- lofacialregion,therehavebeennumerous discussions on the effectiveness of conductingpreoperativevirtualoperations from clinical imaging data (e.g., CT)16. FortheLeFortIosteotomyprocedurein particular,theapplicationof3Dprinting anduseofpatient-specificimplants(PSIs) producedby3Dmetalprintinghavebeen utilized to determine the maxilla posi- tion15,17.PSIsproducedby3Dmetalprint- inghavenotyetbeenclinicallyapproved inJapan; thus, 3D printers thatuse bio- compatiblematerialshavebecomemain- streamforsurgicalguides,similartowhat isoutlinedinthecurrentreport.
Additionally, image-guided surgery (IGS)hasbeenreported,where3Drecon- Fig.6. SignedvaluedeviationbetweenTvandT1.Theaccuracyateachpointwascalculated
withregardtotheXYZaxes.Positivevaluesrepresentunder-correctionandnegativevalues representover-correctionoftheplannedmaxillaryposition.
structionsoftheclinicalimagingdataare made;anatomicalstructuressuchasblood vessels,nerves,andbonesareconfirmed, andthe spatialrelationshipsbetweenthe surgical tools and patients can be con- firmed inreal-timewithsurgicalnaviga-
tion9. Arecentstudydiscussedacasein whichwaferlessnavigationbasedonaug- mentedreality(AR)wasappliedasanIGS technique to a Le Fort I osteotomy procedure18. However, IGS guides the maxillary repositioningand confirmation
throughamonitorandisthustwo-dimen- sionalinnature.
The Microsoft HoloLens has recently beenappliedtosurgicalnavigation,how- ever almostnoapplications ofthis tech- nology in thefield oforal surgery have Fig.7. SignedvaluedeviationbetweenTvandT1ateachpointontheXYZaxes.Anerrorof+3.93mmwasobservedforANSontheX-axis,and anerrorof+1.98mmforANSand3.24mmforPNSwasobservedontheY-axis.
beennoted.SurgeonsusingtheMicrosoft HoloLenscanmaintainthe3Dhologram reconstructed fromclinicalimaging data ontheoperatingfieldwithoutbreakingeye contact and can conduct surgery while continuing tolookat thepatient19.Stan- dard AR and navigation systems have limitations regardingportability, calibra- tion,andtracking,andtheircostscanbe prohibitively high. In contrast, MR- HMDs, suchastheMicrosoftHoloLens, are easy to carry, project an accurate hologram in space, and operations can be conducted through gestural move- ments. Additionally, sanitary operations can beconductedduetothegesturalna- tureoftheprocedure.Asseenabove,the limitations imposed during surgery are graduallydisappearing20.
Up to now, only a few studies have includedacontrolgroup,andmanyhave comparedthepreoperative virtualopera- tion3Dimage andthepostoperativeCT image2,11,19.Oneoftheclinicalgoalsisto achieve an error of less than 2mm be- tweenthevirtualoperation3Dimageand the postoperative CT image21. In this study, the analysisshowed a3D surface accuracy ranging from79.9% to97.1%, with an average of 90.3%. Previous reportsofLeFortIosteotomiesperformed
using image-based aids or a 3D-printed metal PSI have reported bone surface alignment errors <2mm of between 62% and 100%, with an average of 83.8%to92.7%2,9.Thesevaluesindicate thatthemethodusedinthepresentstudy wasabletoreproducethemaxillaposition withhighaccuracy.
Thepoint-basedanalysisofthecurrent applicationshowedthatthemedianerror foreachaxiswhencomparingTvandT1
was <2mm, and nostatistically signifi-
cant differences were observed between TvandT1foranyofthepointsonanyof the axes. The error in positioning the maxillausinga wafermade fromacon- ventionalcastmodelwasapproximately5 mm22.Thus,themaxillawasconsideredto havebeenrepositionedasspecifiedbythe preoperativevirtualoperation.
Regardingthedirectionofmovementof themaxilla,highaccuracywasobservedin thetiltimprovementcases(Table1).Similar topreviousreports, aloweraccuracywas observedinspecifictiltimprovementcases, suchasinbonerepositioningintheimpac- tiondirection(Table1),andinverticalrepo- sitioningofthePNS(Table1)11.Itisthought thatinthesecases,itwasdifficulttotrimthe obstructed area, such as the area in the vicinity of thedescendingpalatine artery,
whenrepositioningthemaxillaasspecified bythe virtualoperation,evenwithCAD/
CAMusage;impactioncannotoccurinthe expectedlocationduringmaxillareposition- ing. As a result, it is thought that when comparing each axis, significant Y-axis differences were observed in the signedvaluebecauseunder-correctionwas morelikelytooccurintheY-axisthaninthe otheraxes.Insomecases,negativevaluesin theX-axisdirectionforANSand A-point werethoughttobepossibly dueto infiltration intheANSsectionfrommucosalabrasion during surgery, or the presence of com- pressedbone resorptionduetodrilling in theANSfollowingalarbasecinchsuturesto controlalarexpansion,anditbeingligated withan absorbable suture. Therefore, the accuracyoftheabsolutevaluetendstobe lower for ANS on the X-axis, and it is possiblethatasignificantdifferenceoccurs intheabsolutevaluewhencomparingtheY andZaxes;thus,weconsideredthattheANS tendedtobeinfluencedasameasurement iteminthisstudy.
The HoloLens was used to confirm reproducibilityaftermaxillarepositioning inthestudymethod;however,theabove results were likely due to the fact that although patient/hologram alignment errorscanbedetectedfairlyeasily,includ- ingtiltimprovementcases whereconfir- mationcanbe done fromthefront side, PNS repositioning confirmation is con- ductedfromthelateralside.Recentstud- ieshavecalculatedthaterrorsinphysical object/hologram alignment due to the HoloLens are on average 2.5mm23 or 2.77mm24, and it is currently believed thatthedeterminationofmaxillaposition usingonlyHMDs isdifficult. Soguided surgeryusingonlyMRisideal,butinthis study,MRwasusedtoimprovetheaccu- racy ofCAD/CAM surgical guides. For thisreason,wesoughttoincreaseaccura- cybyconfirmingin3Dandbythesym- metryoftheLeFortIsegmentduringthe operation that the maxilla was reposi- tioned as specified in the preoperative ANS,anteriornasalspine;PNS,posteriornasalspine.
Forallpoints,themedianerrorbetweenthepreoperativevirtualoperationandthe1-monthpostoperativeCTdatawaslessthan2mm,withno significantdifferences.Hence,itisconsideredthatthesurgerywasperformedwithanaccuracyclosetothepreoperativevirtualoperation.
Fig.8. AbsolutevaluedeviationbetweenTvandT1.Theaccuracyateachpointwithregardto theXYZaxeswascalculated.Themedianabsolutevalueforeachaxiswasasfollows:0.56mm (range0–3.93mm)ontheX-axis,0.29mm(range0.01–3.24mm)ontheY-axis,and0.33mm (range0–2.59mm)ontheZ-axis.
virtualoperationbycombiningtheHolo- Lens withCAD/CAM-fabricated splints.
Therefore, the accuracy is further im- provedbyusingMR.Thismethodappears to be novel as it can be used to check intraoperatively using MR that the
planned maxillary positioning (as deter- minedfromthevirtualoperation)canbe performedusingtheCAD/CAMmanufac- tureddevices.
In this study, the preoperative virtual operationwasrecreatedwithhighaccura-
cy by combining MR and CAD/CAM techniques.Comparedtopreviousstudies, the 3Dsurface andpoint-based analyses indicate thatthismethodwas abletore- producethemaxillarypositionwithahigh levelofaccuracyonallaxes.Additionally, Fig.9. AbsolutevaluedeviationbetweenTvandT1ateachpointontheXYZaxes.Similartothesignedvalues,theaccuracytendedtobelowerfor ANSontheX-axisandPNSontheY-axis.
Notapplicable.
Competinginterests
Theauthorsdeclarenoconflictofinterest.
Ethicalapproval
The study was approved by the Ethics Committee of Tokyo Dental College, Tokyo,Japan(No.794844).
Patientconsent
Writtenpatientconsentwasobtained.
Acknowledgements. We wish to thank HoloeyesInc.fortechnicalsupportduring the creation of the MR application and Meishin ElectricCo., Inc.fortheappro- priatedesignofthedevices.
References
1. LindorfHH,Steinha¨userEW.Correctionof jaw deformities involving simultaneous osteotomyof themandibleand maxilla.J MaxillofacSurg1978;6:239–44.
2. MazzoniS,BianchiA,SchiaritiG,Badiali G,MarchettiC.Computer-aideddesignand computer-aided manufacturing cutting guides andcustomizedtitanium platesare usefulinuppermaxillawaferlessreposition- ing.JOralMaxillofacSurg2015;73:701–7.
http://dx.doi.org/10.1016/j.
joms.2014.10.028.
3. SharifiA,JonesR,AyoubA,MoosK,Walker F,KhambayB,McHughS.Howaccurateis modelplanning fororthognathicsurgery?IntJ Oral Maxillofac Surg 2008;37:1089–93.
http://dx.doi.org/10.1016/j.ijom.2008.
06.011.
4. TominagaK,YoshiokaI,KhanalA,FurutaN, HabuM,FukudaJ.Asimplemethodforbone positioning of mandibular segments.Int J Oral Maxillofac Surg2006;35:856–60. http://dx.
doi.org/10.1016/j.ijom.2006.03.004.
5. WongKC,KumtaSM,SzeKY,WongCM.
Useofapatient-specificCAD/CAMsurgical jigin extremitybone tumorresectionand custom prosthetic reconstruction. Comput Aid Surg 2012;17:284–93. http://dx.doi.
org/10.3109/10929088.2012.725771.
6. Park JH,Yong BL, SangYK, HyungJK, Young SJ,HwiDJ. Accuracyofmodified CAD/CAM generated wafer for orthog-
2018;2:2.
017-0033-2.
8. XiaJJ,GatenoJ,TeichgraeberJF,Christen- senAM,LaskyRE,LemoineJJ,Liebschner MA.Accuracyofthecomputer-aidedsurgi- calsimulation(CASS)systeminthetreat- ment of patients with complex craniomaxillofacialdeformity:apilotstudy.
J Oral Maxillofac Surg 2007;65:248–54.
http://dx.doi.org/10.1016/j.
joms.2006.10.005.
9. BadialiG,RoncariA,BianchiA,TaddeiF, Marchetti C, Schileo E. Navigation in orthognathic surgery: 3Daccuracy.Facial Plast Surg 2015;31:463–73. http://dx.doi.
org/10.1055/s-0035-1564716.
10. ParkSH,YuHS,KimKD,LeeKJ,BaikHS.
Aproposalforanewanalysisofcraniofacial morphologyby3-dimensionalcomputedto- mography.AmJOrthodDentofacialOrthop 2006;129:600.e23–e. http://dx.doi.org/
10.1016/j.ajodo.2005.11.032.
11. HeufelderM,WildeF,PietzkaS,MaschaF, Winter K,Schramm A,RanaM. Clinical accuracyofwaferlessmaxillarypositioning usingcustomizedsurgicalguidesandpatient specific osteosynthesis in bimaxillary orthognathic surgery. J Craniomaxillofac Surg 2017;45:1578–85. http://dx.doi.org/
10.1016/j.jcms.2017.06.027.
12. OngTK,BanksRJ,HildrethAJ. Surgical accuracyinLeFortImaxillaryosteotomies.
BrJOralMaxillofacSurg2001;39:96–102.
http://dx.doi.org/10.1054/bjom.2000.0577.
13. MazzoniS,BadialiG,LancellottiL,Babbi L,BianchiA,MarchettiC.Simulation-guid- ed navigation:a newapproachtoimprove intraoperative three-dimensional reproduc- ibilityduring orthognathicsurgery. JCra- niofac Surg 2010;21:1698–705. http://dx.
doi.org/10.1097/SCS.0b013e3181f3c6a8.
14. MarchettiC,BianchiA,BassiM,GoriR, LambertiC,SartiA.Mathematicalmodeling and numericalsimulationinmaxillo-facial virtual surgery(VISU). J CraniofacSurg 2006;17:661–7. http://dx.doi.org/10.1097/
00001665-200607000-00009.
15. KimJW,KimJC,JeongCG,CheonKJ,Cho SW,ParkIY,YangBE.Theaccuracyand stability of the maxillary position after orthognathicsurgeryusinganovelcomput- er-aided surgical simulationsystem. BMC Oral Health 2019;19:18. http://dx.doi.org/
10.1186/s12903-019-0711-y.
16. Keeve E, Girod S, Girod B, Deformable modeling of facial tissue for craniofacial surgery simulation.Ho¨hne KH, Kikinis R, editors.Visualizationinbiomedicalcomput-
ThammOC,RothamelD,Zo¨llerJE.Com- puter-assistedorthognathicsurgery: wafer- less maxillary positioning, versatility, and accuracy of an image-guidedvisualisation display. Br J Oral Maxillofac Surg 2013;51:827–33. http://dx.doi.org/10.1016/
j.bjoms.2013.06.014.
19.WitowskiJ,DarochaS,Kownacki,PietrasikA, PieturaR,BanaszkiewiczM,KaminskiJ,Bie- dermanA,TorbickiA,KurzynaM.Augmented realityandthree-dimensionalprintinginpercu- taneousinterventionsonpulmonaryarteries.
Quant Imaging Med Surg 2019;9:23–9.
http://dx.doi.org/10.21037/qims.2018.09.08.
20.Kress BC, Cummings WJ. 11-1: Invited Paper: Towardstheultimate mixedreality experience: HoloLens display architecture choices.SIDSymposiumDigestofTechnical Papers 2017:127–31. http://dx.doi.org/
10.1002/sdtp.11586.48.
21.StokbroK,AagaardE,TorkovP,BellRB, ThygesenT.Virtualplanninginorthognathic surgery. Int J Oral Maxillofac Surg 2014;43:957–65. http://dx.doi.org/10.1016/
j.ijom.2014.03.011.
22.EllisE.Accuracyofmodelsurgery:evalua- tionofanoldtechniqueandintroductionofa new one. J Oral Maxillofac Surg 1990;48:1161–7. http://dx.doi.org/10.1016/
0278-2391(90)90532-7.
23.Gibby JT,SwensonSA,CvetkoS,RaoR, JavanR.Head-mounteddisplayaugmented realitytoguidepediclescrewplacementuti- lizingcomputedtomography.IntJComput AssistRadiolSurg2018;14:525–35.http://dx.
doi.org/10.1007/s11548-018-1814-7.
24.LiebmannF,RonerS,vonAtzigenM,Scar- amuzzaD,SutterR,SnedekerJ,FarshadM, Fu¨rnstahlP.Pediclescrewnavigationusing surfacedigitizationontheMicrosoftHolo- Lens. Int J Comput Assist Radiol Surg 2019;14:1157–65. http://dx.doi.org/
10.1007/s11548-019-01973-7.
Address:
MasahideKoyachi
DepartmentofOralPathobiologicalScience andSurgery
TokyoDentalCollege 2-9-18Misaki-cho Chiyoda-ku 101-0061Tokyo Japan
Tel.:+81363809246;
Fax:+81332623213 E-mail:[email protected]
