Page 59 - Demo
P. 59


                                    Cranialshapeevaluation573SUPERIMPOSITIONOF3DPHOTOGRAPHSForadequateanalysisevery3Dphotographneededtobepositionedinthesamecoordinate system and therefore rotations and translations needed to be eliminatedToachieve thisall3Dphotographswereimportedinto the3DMDPatientsoftware(3dMDAtlantaGAUSA)Eightanatomicallandmarksweremanuallyidentifiedonthe3Dphotographs(Table1)The3DphotographsandlandmarksweresubsequentlyexportedtothetechnicalcomputingsoftwareMatlab(MATLABv2012b TheMathworksIncNatickMAUSA)InMatlab ahorizontalplane definedbytheexocanthionendocantionandpretragionpointsandaverticalplaneperpendiculartothehorizontalplaneandpositionedthroughthesubnasalpointweregeneratedBasedonthemethoddescribedbyBronsetaltheplaneswereusedtoeliminatetherotations9(Figure1)All3Dphotographsweretranslatedtotheoriginofthereferenceframeusingthecomputedcranialfocalpoint(CCFP)presentedbydeJongetal10 Thisresultedinadatasetof3DstereophotographsinthesamecoordinatesystemwithoutrotationalandtranslationaldifferencesRESAMPLINGANDSCALINGAllthealignedandcorrectlypositioned3Dphotographsconsistedofadifferentnumberofvertices(3Dpoints)Forfurtheranalysisofthe3Dphotographsitwasrequiredtocreatemeshdatawiththesamenumberofverticesandthereforearesampling wasappliedResampling wasdoneusingareferenceshapeconsistingofahemi-icosphereintheformofacranialshapeandaraycastingalgorithm11Atotalof4501rayscastfromtheoriginofthereferenceshapewereusedandthereforeeverycranialshapeconsistedof4501verticesafterresamplingThevolumeofeverycranialshapewascalculatedusingthehorizontalplaneasalowerboundaryFinallyeverycranialshapewasscaledtowardsthesamevolumetoeliminatedifferencesincranialsizeAscalingidenticalineverydirectionwasusedwhichadjustedonlythesizeandpreservedtheshapeofthemodified3DphotographsPRINCIPALCOMPONENTANALYSISPCAisastatisticalmethod whichcandescribealargenumberofhighdimensionaldatawithasmallernumberofrelevantparameters12–143DphotographsofapopulationcanbeanalysedbyPCAtofindandisolatethedifferentvariationsofthe3Dshapesin thispopulationPCAcanbeused toextractonly therelevantandmostprominentvariationssuchasthelengthoftheskullanddisregardminor
                                
   53   54   55   56   57   58   59   60   61   62   63