LocalizationofectopicmaxillarycaninesisCBCTmoreaccuratethanconventionalhorizontalorverticalparallax?PaulS.Serrant,GrantT.McIntyreandDonaldJ.ThomsonDepartmentofOrthodontics,NHSTayside,Dundee,UKObjective: To compare the accuracy of cone-beam CT (CBCT) with conventional horizontal and vertical parallax forthelocalizationofectopicmaxillarycanines.Design:Observationalstudy.Methods:Atypodontwasconstructedusinghumanteethembeddedinradio-denseimpressionmaterial, containingasimulatedectopiccanine. Imagesets(vertical/horizontal parallaxandCBCTscans)wereobtainedforninepositions. Aphotographconfirmedthetrue position of the canine. Six observers used horizontal (HP) and vertical (VP) parallax techniques withconventional radiographsandtheaxial, coronal andsagittal CBCTviewstolocatethecaninetip. Imagesweremanipulated as required using proprietary software (http://www.carestream.com and http://www.i-cat.com). Intra-observerreproducibilitywascalculatedbyoneobserverre-evaluatingtheimagesetsafter2months. Individualobserver validitywas calculatedinrelationtothephotographicpositionofthecanineusingweightedKappa.Differences in the proportion of correct locations between CBCT and vertical/horizontal parallax were tested usingMcNemar tests (P,0.05). Results: Intra- andinter-observer agreements wereexcellent (0.8985) andsubstantial(0.7528), respectively. Individual observer validity was substantial-excellent (0.73680.900). The canine position wascorrectly identified in 94% of cases located using CBCT, 83% using HP and 65% using VP. The differences betweenCBCTandverticalandhorizontal parallaxwerehighlystatisticallysignificant(P,0.01). Conclusion: CBCTismoreaccuratethaneitherhorizontalorverticalparallaxforthelocalizationofectopicmaxillarycanineteeth.Keywords:CBCT,horizontalparallax,impactedcanine,verticalparallaxReceived 10 May 2013; accepted 24 June 2013IntroductionTheincidenceofimpactedpermanentmaxillarycaninesvaries from 0.92 to 2.2%, depending upon the populationstudied (Hunter, 1983). Of these, 85% are located palataltothedental archandtheremainder areeither inthelineof thearchorbuccal (Ortonet al., 1995). Palatalimpactions arethought toresult fromeither alackofguidance by the distal surface of the lateral incisor, or dueto a genetic predisposition. Buccal impactions and thoseinthelineofthearchresultfromadeciencyofspacewithin the dental arch and may be the result of aninherent crowding tendency or due to early loss ofprimary teeth with loss of arch length (Orton et al., 1995).The parallax technique was rst describedby Clark(1910) using two periapical radiographs exposed atdifferent angles in the horizontal plane. In 1952,Richards (1952) introducedthe concept of the verticaltube shift. Keur (1986) suggested that the use of alargerocclusallmallowedforfurthermovementoftheX-ray tube, whichresults inclearer distinctions tobemade for ectopic teeth that are close to the line of the arch.Horizontal parallax has been found to be moreaccurate for the localization of ectopic maxillary caninesbyArmstrongetal. (2003)Theyfoundthatthereweresignicantlymore unsure diagnoses withthe verticalparallax technique, and suggested that horizontalparallaxismoreaccurateduetoavailabilityof cleareranatomicalreferencepointsinthehorizontalplane.The magnicationtechnique, describedby Chaushuet al. (1999), uses amagnicationindex(calculatedasa ratio) applied to a dental panoramic tomogramincombination with vertical restriction criteria (three bandsof zoning; coronal, mid and apical) based on the positionoftheectopiccaninerelativetotheipsilateralmaxillarycentral incisor. While the bucco-palatal position of mostectopic canines positionedinthe middle/palatal zones(relative to the ipsilateral central incisor) can be correctlyidentied with this technique, ectopic canines lying in theSCIENTIFICSECTION JournalofOrthodontics,Vol.41,2014,1318Addressforcorrespondence:G.McIntyre,DepartmentofOrthodontics, NHS Tayside, 2 Park Place, Dundee DD1 4HR, UKEmail:grant.mcintyre@nhs.net#2014BritishOrthodonticSociety DOI10.1179/1465313313Y.0000000076apical zonearenotreliablylocatedwiththistechnique(Sudhakaret al.,2009).Cone-beam CT (CBCT) technology is considered to bethe most accurate method for the localization of impactedteeth (SEDENTEXCT Project, 2011); however, it isassociated with a higher overall effective dose thanconventional radiography. TheSEDENTEXCTproject20082011 (http://www.sedentexct.eu) produced evi-dence-based guidelines for the use of CBCT examinationsindental andmaxillofacial imaging, whichincludethelocalization of ectopic maxillary canines.Arecent review(Rossini et al., 2012) suggestedthatCBCTandhorizontal/vertical(HP/VP)imagesetsmayproducedifferent resultsandnotedalackof evidenceregardingthediagnosticaccuracyandeffectivenessofCBCTinthelocalizationof maxillaryimpactedteeth.Haney et al. (2010) and Botticelli et al. (2011) evaluatedthe potential for CBCT, HP and VP to accurately locateectopic maxillary canines, but did not compare thelocationwiththat of the true positionof the ectopictooth(Table1). Alqerbanet al. (2011) comparedtheuse of the magnication technique utilizing a singlepanoramic radiographwithtwodifferent CBCTsys-tems. Theyevaluatedthe locationandorientationofthe ectopic canine, and the presence, severity andlocation of lateral incisor resorption nding statisticallysignicantgreateragreementbetweenobserversforallvariableswhenusingCBCTcomparedtothemagni-cationtechnique,but nosignicantdifferences betweentheCBCTmachines.NullhypothesisCBCTissimilartoeitherHPorVPfortheaccuracyoflocalizationofectopicmaxillarycanineteeth(P,0.05).AimTo compare the accuracy of CBCT, conventionalhorizontal and vertical parallax techniques in thelocalizationofectopicmaxillarycanineteeth.Materials andmethodsResearch and development approval and study sponsor-ship were obtained from Tayside Medical ScienceCentre.Ethicalapprovalwasnotrequired.Atypodontwithasimulatedmaxillarydentalarchwasconstructedutilizing a radio-dense poly-vinylsiloxane impressionmaterial withembeddedextractedhumanteeth, whichsimulated an ectopic, impacted maxillary left caninetooth (Figure1). As the human teeth used in thisexperiment were donated for dental research anon-ymouslybeforeSeptember 2003havingbeenextractedas part of routine dental treatment, this study wascompliant with the Human Tissue Act 2004. Thecanine was secured in wax and the position was adjustedto produce nine simulated absolute positions (threebuccal, threepalatal andthreeinthelineof thearch)at three vertical levels relative to the adjacent late-ral incisor tooth (high, medium and low) (Table2)(Figure1). Aphotographwastakenbeforeeachradio-graph set to conrm the actual location of the crown tipof the canine.A CBCT scan [I-CAT next generation, ImagingSciences International, Hateld, PA, USA (http://www.i-cat.com)] alongwithvertical parallaxandhor-izontal parallaximagesets[SironaxG5, SironaQdentPhosphor Plate System; Sirona Dental GmbH, Salzburg,Austria(http://www.sirona.com)]wasobtainedforeachof the nine simulated ectopic canines, producing 27standardized image sets (Figure2). The typodont mount-ingpositions for thedental panoramictomogramandCBCT scans were standardized. For the CBCT scans, theeld of view was 6cm, scan time was 8.9s and voxel sizewas0.3mm3.Aputtyindexwasusedtolocatethelmholderfortheperiapicalimagesandtheocclusal radio-graph was taken using a standard angulationSix observers (one consultant in Orthodontics, oneconsultant in Dental and Maxillofacial Radiography,three specialists inOrthodontics andone FoundationDentist) evaluatedthe 27image sets ina single-blindrandomized order (a random sequence was generated foreachobserverusinghttp://www.random.org).Observerswereabletomanipulateimagesusingproprietarysoft-ware (I-CATvision, http://www.i-cat.com; CarestreamPACS, http://www.carestream.com) as necessary and thelocation of the cusp tip (palatal, in line of arch or buccal)for eachof the27imagesets was notedusingVP/HPparallax where appropriate using a data collection sheet.The data were transferred fromthe proforma to aspreadsheet (Microsoft Excel, Redmond, CA, USA)(Table3) andthedifferences betweenthe image typesweredetermined. Oneobserver re-evaluatedtheimagesets 2months later.Figure1 Puttyindexforperiapical lms14 Serrantetal. ScientificSection JOMarch2014StatisticalanalysisIntra-observer reproducibility (for observer 1) and inter-observer reproducibility (for all six observers) weredetermined using Cohens weighted kappa (Cohen, 1960)andusingFleisssweightedkappa(Fleiss,1981),respec-tively. Individual observer validity (comparison of indivi-dual results withthe position of eachindex toothasdetermined with the photograph) was assessed usingCohens weighted Kappa. The Kappa values wereclassied according to Landis and Koch (1977) as follows:,0asnoagreement, and00.20asslight, 0.210.40asfair,0.410.60asmoderate,0.610.80assubstantialand0.811 as excellent agreement. The proportion of correctlylocalizedectopiccanineswasdeterminedandthediffer-ences betweenCBCTandVP/HPparallaxwere testedusing McNemar tests (P,0.05).Figure2 (a) Periapical radiographs for horizontal parallax; (b) anterior occlusal (75u) and dental panoramic tomogram(DPT) used for vertical parallax; (c) screenshot of a CBCT scan imaging (coronal and sagittal views with 3D reconstruction)JOMarch2014 ScientificSection EctopiccaninesisCBCTbetterthanparallax? 15ResultsIntra-observer reproducibilitywas excellent (K50.8985)and inter-observer agreement was substantial (K50.7528).Individual observer validity was substantial for oneobserver (K50.7368) and excellent for the other ve(K50.82350.900) observers (Table4).The location of the ectopic canine was correct for 94%of occasions whenusingCBCT, 83%whenusingHPand 65% when using VP and the differences were highlystatistically signicant (P,0.01). The null hypothesiswas therefore not supported. Two cases where theectopicmaxillarycaninewaspositionedbuccallytothelineof thearchwereclassiedas beinginthelineofthe arch by all observers using VP (Table3). There wereno cases where a buccal canine was judged to be palatalorviceversa.DiscussionIn this study, CBCTwas more accurate than eitherhorizontal or vertical parallax for the localizationofectopic maxillary canine teeth. The proportion ofcorrectlyclassiedlocationswasthehighestwhenusingCBCT(94%); this was lower whenHPwas employed(83%) and was lowest with the VP technique (65%). Twocases where the maxillary canine was positionedbuccallytothelineofthearchwereclassiedwronglybyall observersusingVP, but werecorrectlylocalizedwhenusingCBCT. Therewasatendencyforbuccallypositioned ectopic canines to be incorrectly located withthe VP technique. The poorer accuracy of the VPtechnique in comparison to HP is in line with the resultsby Armstrongetal.(2003) andMasonet al. (2001)Ourproportion of correctly classied images when using HPandVPwere broadly similar tothose by Armstronget al. (2003), who found signicantly more unsurediagnoses with the VP technique. Poorer accuracy whenusing VP may arise due to foreshortening of the teeth onanocclusal radiographandwheretheectopiccanineisclosetothelineof thearch, thisprojectionerrormayresult inconfounding. UnlikeArmstronget al. (2003),we didnot allowour observers toassignanunsuregrade.ThendingthatlocalizationusingCBCTimageswassignicantlymoreaccurate thaneither HPorVPin ourstudy differs to the ndings of the study by Haney et al.(2010) They found that the use of traditional 2Dradiographs (collectivelyHPandVP) weremarginallysuperior to CBCT. The difference between the studies isexplainablebecauseHaneyet al. (2010) providedtheirobservers with hard copy print-outs of the CBCTimages,whereasourobserverswereabletomanipulatetheCBCTimagesasmuchasrequiredbytheobserver Table1PreviousstudiescomparingCBCTand2DradiographsStudyGenderObservers2DimagesCBCTimagesOutcomesConclusionHaneyetal.(2011)186M12F4orthodontists,3oralsurgeons:4with,10-yearsexperienceand3with.10-yearexperienceDPT/occlusal(VP)2periapicals(HP)3Dimagesonglossyphotopaper:anterior,posterior,rostral-caudal,caudal-rostral,labial,andpalatalviewsLocation,resorption,treatmentplanning,spontaneouseruption,furtherresorption,additionalimagesrequiredNostatisticallysignificantdifferenceinaccuracyoflocationofectopiccanineswith2Dor3DimagesBotticellietal.(2011)2710M17F8dentists:3inexperiencedorthodontictrainees,2moderately-experiencedorthodontictraineesand3specialistsorthodonticsDPT,lateralheadfilm,andavailableperiapicalsSeriesofstaticimages:683DMIP/volumerenderingreconstructions1220axialimages1curvedplanarreformationSeriesofmultiplanarreconstructions2obliqueplanarreformationrecordingthebucco-lingualandmesio-distaldimensionofthetoothPositionofthecanine,rootresorptionaffectinglateralincisor,evaluationofdifficultyoftreatmentandqualityoftheimages,determinationoftreatmentstrategy3Dimagesalloweddeterminationofthemesio-distalpositionofthecanineapex,whilethebucco-palatalpositioncouldbeassessedwithlessuncertainty.Theoverlapwiththelateralincisorandpresenceofrootresorptionwereperceivedasmoresevereafter3Devaluation.Greatercasedifficultyanddifferencesintreatmentplanningwereidentifiedwith3Dimaging16 Serrantetal. ScientificSection JOMarch2014using proprietary software. This mirrors clinical practice.Furthermore, unlikeHaneyet al. (2010) andBotticelliet al. (2011), we identied the location of the index toothusingaphotograph before imagingtocomparethedatawiththeexactlocation. Theseinvestigationsalsoreliedon the congruence of their observers as the primaryoutcome measure. Botticelli et al. (2011) found that inter-observeragreement wasbetterwhenusingCBCTthan2Dimages, althoughtheycouldnot conrmthe truelocation of the canine and their results need to be viewedin relation to this source of bias.In this study, which compared the accuracy of CBCT,HPand VPfor the localization of ectopic maxillarycanines, intra-observer reproducibility and inter-observeragreement were excellent (K50.8985) and substantial(K50.7528), respectively. Individual observer validitywas substantial for one observer (K50.7368) andexcellentfortheotherve(K50.82350.900)observers(Table2), conrming consistency of observers with arange of clinical expertise. This was an ex vivo study andas such, the ectopic maxillary canine was simulated.However, the process for imaging each position wasstandardizedandthe images were assessedinrandomorder to minimize any bias. As grossly displaced buccalypositionedectopicmaxillarycaninescanbelocatedbydigital palpationduringclinical examination, thisstudyTable3 ResultsforallsixobserversImageset Correctlocation Imagingtechnique Observer1 Observer2 Observer3 Observer4 Observer5 Observer61 P CBCT P P P P P P2 A Vertical B B B B B B3 B CBCT B A B B B B4 B Vertical B B B B B B5 A Vertical A A A A A A6 P Vertical P A A A A A7 P Vertical P P P P P P8 A CBCT A A A A A A9 A Horizontal B A A A A A10 B Horizontal B B B B B B11 P Vertical P P A P P P12 A CBCT A A A A A P13 A Horizontal B B A A B A14 A CBCT A A A A A A15 P Horizontal P P P P P P16 A Vertical B B B B B B17 P Horizontal P P A P P P18 B CBCT B A B B B B19 B Vertical B B B B A B20 P CBCT P P P P P P21 A Horizontal A A A A B A22 B CBCT B B B B B B23 B Horizontal B B B A B B24 B Horizontal B B B B B B25 P Horizontal P P A P A P26 P CBCT P P P P P P27 B Vertical B B B B B BTable2 PositionofectopicmaxillarycanineandlocationofcrowntipVerticalandhorizontalpositionofcanine LocationofcaninecrowntipVerticalheightofectopiccanines NHigh:crowntippositionedatthesameleveloftheapexoftheadjacentleftlateralincisortoothNMedium:crowntippositionedataverticalheight5mmocclusaltotheapexoftheadjacentleftlateralincisortoothNLow:crowntippositionedataverticalheight10mmocclusaltotheapexoftheadjacentleftlateralincisortoothHorizontal(bucco-palatal)displacementofectopiccaninesNPalatal:crowntippositionedapproximately25mmpalatallytothelineofbestfit,correlatingtothecurvatureofthesimulateddentalarchNInlineofarch:crowntippositionedascloseaspossibletothelineofbestfit,correlatingtothecurvatureofthesimulateddentalarch.NBuccal:crowntippositionedapproximately25mmbuccallytothelineofbestfit,correlatingtothecurvatureofthesimulateddentalarchJOMarch2014 ScientificSection EctopiccaninesisCBCTbetterthanparallax? 17only assessed minimally displaced (25mm) ectopicmaxillary canines, where radiographic examinationwould be essential.CurrentguidancestatesthatCBCTmaybeindicatedfor the localized assessment of an impacted tooth(including consideration of resorption of anadjacenttooth) (The SEDENTEXCT Project, 2011). WhileCBCThasbeenshowninthisstudytobesuperiortoHP/VP, the smallest volume size compatible withtheclinical situation should be selected to minimize theradiation dose for the patient where there is anysuspicionofresorptionofanadjacenttoothroot.ConclusionCBCT is more accurate than either horizontal or verticalparallaxforthelocalizationofectopicmaxillarycanineteeth.AcknowledgementsMrsAlisonMenhinick(Superintendent Radiographer)and Mr Simon Scott (Medical Photographer) kindlyrecordedalltheimagesforthisstudy.ReferencesAlqerbanA, JacobsR, FieuwsS, WillemsG. Comparisonof twoconebeamcomputedtomographicsystemversuspanoramicimagingforlocalizationof impactedmaxillarycanines anddetectionof root resorption. Eur JOrthod2011;33:93102.Armstrong C, Johnston C, Burden D, Stevenson M. Localizing ectopic maxillarycanineshorizontalorverticalparallax?EurJOrthod2003;25:585589.Botticelli S, VernaC, CattaneoPM, HeidmannJ, MelsenB. Twoversusthreedimensional imaginginsubjectswithuneruptedmaxillarycanines. EurJOrthod2011;33:344349.CohenJ.Acoefficientofagreementfor nominalscales.EducPsychMeas1960;20:3746.FleissJL.StatisticalMethodsforRatesandProportions.2ndedn.WileySeriesin probability and mathematical statistics. New York: Wiley, John & Sons,1981.Chapter13,pp.212236.Chaushu S, Chaushu G, Becker A. The use of panoramic radiographs to localisedisplacedmaxillarycanines.OralSurgOralMedOralPatholOralRadiolEndod1999;88:511516.ClarkC.Amethodofascertainingthepositionofuneruptedteethbymeansoffilmradiographs.ProcRSocMed1909;3:8790.Haney E, Gansky SA, Lee JS, Johnson E, Maki K, Miller AJ,et al. Comparative analysis of traditional radiographs and cone-beamcomputedtomographyvolumetricimagesinthediagnosisandtreatmentplanningofmaxillaryimpactedcanines.AmJOrthodDentofacialOrthop2010;137:590597.HunterSB. Treatmentoftheuneruptedmaxillarycanine.Part1Preliminaryconsiderationsandsurgicalmethods.BrDentJ1983;154:294296.KeurJJ.Radiographiclocalizationtechniques.AustDentJ1986;31:8690.OrtonHS, Garvey MT,PearsonMH.Extrusionof the ectopic maxillary canineusing a lower removable appliance. Am J Orthod Dentofacial Orthop 1995;107:349359.LandisJR,KochGG.Themeasurementofobserveragreementforcategoricaldata.Biometrics1977;33:159174.MasonC, PapadakouP, RobertsGJ. Theradiographiclocalizationofectopicmaxillary canines: a comparison of methods. Eur J Orthod 2001; 23: 2534.RichardsAG. Roentgenographiclocalizationof themandibularcanal. JOralSurg1952;10:325329.Rossini G, Cavallini C, CassettaM, GalluccioG, BarbatoE. Localizationofimpacted maxillary canines using cone beamcomputed tomography:reviewoftheliterature.MSAnnStomatol(Roma)2012;3:1418.Sudhakar S, Patil K, Mahima VG. Localisation of impacted permanentmaxillarycanineusingsinglepanoramicradiograph. IndianJDent Res2009;20:340345.TheSEDENTEXCTProject.RadiationProtection:ConeBeamCTforDentalandMaxillofacial Radiology. EvidenceBasedGuidelines. Luxembourg:EuropeanCommission,2011.pp.1139.Table4 Individualobservervalidityandintra-observerreproducibilityIndividualobservervalidityIntra-observerreproducibilityObserver1 0.900 K50.8985Observer2 0.833Observer3 0.8235Observer4 0.8889Observer5 0.7368Observer6 0.894718 Serrantetal. ScientificSection JOMarch2014