Computer Vision-Guided Virtual Craniofacial Surgery: A Graph-Theoretic and Statistical Perspective PDF
Preview Computer Vision-Guided Virtual Craniofacial Surgery: A Graph-Theoretic and Statistical Perspective
Advances in Computer Vision and Pattern Recognition Forothertitlespublishedinthisseries,goto www.springer.com/series/4205 Ananda S. Chowdhury (cid:2) Suchendra M. Bhandarkar Computer Vision-Guided Virtual Craniofacial Surgery A Graph-Theoretic and Statistical Perspective Dr.AnandaS.Chowdhury Prof.SuchendraM.Bhandarkar DepartmentofElectronics& DepartmentofComputerScience TelecommunicationEngineering TheUniversityofGeorgia JadavpurUniversity BoydGraduateStudiesResearchCenter 188RajaS.C.MalikRoad Room415 700032Kolkata,WestBengal 30602-7404Athens,GA India USA [email protected] [email protected] SeriesEditors ProfessorSameerSingh,PhD Dr.SingBingKang ResearchSchoolofInformatics MicrosoftResearch LoughboroughUniversity MicrosoftCorporation Loughborough OneMicrosoftWay UK Redmond,WA98052 USA ISSN2191-6586 ISBN978-0-85729-295-7 e-ISBN978-0-85729-296-4 DOI10.1007/978-0-85729-296-4 SpringerLondonDordrechtHeidelbergNewYork BritishLibraryCataloguinginPublicationData AcataloguerecordforthisbookisavailablefromtheBritishLibrary LibraryofCongressControlNumber:2011923551 ©Springer-VerlagLondonLimited2011 Apartfromanyfairdealingforthepurposesofresearchorprivatestudy,orcriticismorreview,asper- mittedundertheCopyright,DesignsandPatentsAct1988,thispublicationmayonlybereproduced, storedortransmitted,inanyformorbyanymeans,withthepriorpermissioninwritingofthepublish- ers,orinthecaseofreprographicreproductioninaccordancewiththetermsoflicensesissuedbythe CopyrightLicensingAgency.Enquiriesconcerningreproductionoutsidethosetermsshouldbesentto thepublishers. Theuseofregisterednames,trademarks,etc.,inthispublicationdoesnotimply,evenintheabsenceofa specificstatement,thatsuchnamesareexemptfromtherelevantlawsandregulationsandthereforefree forgeneraluse. Thepublishermakesnorepresentation,expressorimplied,withregardtotheaccuracyoftheinformation containedinthisbookandcannotacceptanylegalresponsibilityorliabilityforanyerrorsoromissions thatmaybemade. Coverdesign:deblik Printedonacid-freepaper SpringerispartofSpringerScience+BusinessMedia(www.springer.com) To mywifeAnindita – AnandaS.Chowdhury To mywifeSwati,son Pranav,and daughter Asha – SuchendraM. Bhandarkar Foreword Anancientmarinerontheopenseashadtoknowwhereheisbeforehecouldnav- igate to where he had to go. A surgeon, through training and experience, gains a mental image of what that “map” should be. It is called surgical anatomy. This is possiblebecauseforthemostpart,thehumananatomy,atthelarge-scalelevel(1cm to10cm,forexample),isnotexceedinglyvariable.However,toknowtheprecise location of the surgery, dissection is needed to allow either direct visualization or palpationoftheparticularanatomicstructures.Often,suchdissectionisassociated withdisruptionofnormaltissues,interruptionofbloodsupply,prolongingthemag- nitude of surgery, increasing the risk of potential complications, and lengthening thepost-operativeconvalescence.Inrecenttimes,ComputerTomography(CT)and MagneticResonanceImaging(MRI)haveallowedthesurgeontoseethissurgical map at a level of detail and precision not previously possible. Recent advances in both scanning instruments and supporting software have transitioned their impact frommerelyoutsidetheoperatingroomtoinsidethesurgicaltheater,makingintra- operative3Dimagingareality. However, most of the existing intraoperative navigation devices are still bulky, time-consumingtouse,andincreasethepotentialforcontaminatingthesterileop- erativefield.Giventhatthecostperminuteintheoperatingroomcontinuestosky rocket, the more work that can be done before entering the operating room, the moreeffectiveandefficientthesurgeoncanbe.Thismonographdetailsacollabo- rative research endeavor over the past five years at the University of Georgia and theMedicalCollegeofGeorgiaindesigningasoftwaresystemthatcanreconstitute brokenbonesinsilico,eitherthroughagraphicaluserinterfaceorinanautomated fashionfromfracturedetectiontofracturereduction—aprocesswherethedisplaced bone fragments are returned to where they should be. The sizes and shapes of the fracturedbone(s)andtheshapeoftheinsilicoreconstructioncanbeobtainedand thenlinkedtoCAD–CAMdevices.Reconstructionplatescanthenbeshapedtofit precisely the individual anatomy, and the optimal lengths, orientations, and loca- tionsofthevariousscrewscanbedetermined.Armedwiththiscustomizedfixation hardware, a surgeon can drastically reduce the time needed in the operating room and, as long as an accurate fiduciary transfer is made, achieve an accurate reduc- vii viii Foreword tion and fixation of the fractures with significantly less extensive dissection. This approach improves the current state-of-the-art by increasing accuracy, decreasing cost,andreducingboththeoperativetraumaandtheriskofoperativeandpostoper- ativecomplicationstothepatient. Asapracticingplasticsurgeon,IcanrecallmanytimeswhenIwishedIhadsuch asoftwaresystematmydisposal.Itwaslargelyonaccountofthispressingneedthat weembarked,somesixyearsago,uponthisresearchendeavor.Havingtheability todetectandmanipulatethefracturefragments,andreturnthemtotheirpreinjury stateinsilicowillallowthesurgeonstoobtainamoreaccuratereductionandrecon- structionwithlessextensivedissection.Theendresultisamuchshorteroperating time and significantly improved surgical outcome. This monograph is the culmi- natingachievementofthesustainedeffortsofmanypeople,especiallyDr.Ananda Chowdhury,thenaPh.D.candidateattheUniversityofGeorgia,andmydearfriend, Dr. Suchendra Bhandarkar, Ananda’s doctoral research advisor. It is through such collaborative work that we can realize significant, high-impact advances that con- tributetothesurgicaltreatmentofpatientswithcranio-maxillofacialfractures. JackC.Yu,DMD,MD,MSEd,FACS,FADI MilfordB.HatcherProfessorofSurgery andChief SectionofPlasticSurgery SchoolofMedicine MedicalCollegeofGeorgia Augusta,GA,USA Preface Themonographdealswithanimportantapplicationofcomputervisionandpattern recognitionintheareaofmedicalscience,morespecificallyreconstructivecranio- facialsurgery.Craniofacialfracturesareencounteredveryfrequentlyintoday’sfast- pacedsociety;themajorcausesbeinggunshotwounds,motorvehicleaccidents,and sports-relatedinjuries.Surgicalreconstructionischallengingbecausethesurgeons in the operating room have to register the broken bone fragments accurately and under severe time constraints. Within the broad class of craniofacial fractures, the emphasisinthismonographisonmandibularfracturessincethemandibleisoften unprotectedandexposedmakingitespeciallyvulnerabletoaccidentsandinjuries. Atypicalinputtoacomputervision-basedsystemforvirtualcraniofacialsurgeryis asequenceofComputedTomography(CT)imagesofafracturedhumanmandible. ThedetectionoffracturesinCTimages,theotherintegralcomponentofreconstruc- tivecraniofacialsurgery,isoftendifficultbecauseofthecomplexityofthefracture patterns,missingdata,imageintensityinhomogeneities,andpresenceofnoiseand undesiredartifacts. A formal treatment of computer vision-guided craniofacial surgery entails the solvingoftwobroadclassesofproblems,i.e.,computer-aidedfracturedetectionand virtualreconstruction,bothofwhichraiseseveralimportanttheoreticalandpractical issues. From a theoretical standpoint, the monograph discusses several traditional topicsincomputervisionandpatternrecognitionsuchasimageregistration,image reconstruction,combinatorialpatternmatching,anddetectionofsalientpointsand regionsinanimage.Severalusefulalgorithmsandconceptsfromtwotraditionally diverse disciplines, namely, graph theory and statistics are seen to be applicable in this context. The relevant topics from graph theory include maximum-weight graph matching, maximum-cardinality minimum-weight matching for a bipartite graph, maximum-flow minimum-cut determination in a flow graph, and construc- tionofautomorphsofacyclegraph.Themonographdemonstrateshowtheabove graph-theoreticalgorithmscanbeappliedtosolvesomeimportantproblemsincom- putervisionandpatternrecognitionthatpertaintovirtualreconstructivecraniofacial surgery.ThevariousstatisticaltechniquesbroughttobearincludeMarkovrandom fields, hierarchical Bayesian restoration, Gibbs sampling, and Bayesian inference. ix
The list of books you might like
