微博推广营销经验分享.ppt
- 文档编号:2485424
- 上传时间:2022-10-30
- 格式:PPT
- 页数:25
- 大小:970KB
微博推广营销经验分享.ppt
《微博推广营销经验分享.ppt》由会员分享,可在线阅读,更多相关《微博推广营销经验分享.ppt(25页珍藏版)》请在冰豆网上搜索。
EE5359Project-PatternRecognitionDiagnosticusingPhaseOnlyCorrelationtechniquesubmittedbyThejaswiniPurushotham1000616811迈鼎http:
/Fig.1Experimentalsetupformedicalimagediagnostics.Whathasbeendonealready?
Afingerprintmatchingalgorithmusingphaseonlycorrelation1.Theproposedtechniqueisparticularlyeffectiveforverifyinglow-qualityfingerprintimagesthatcouldnotbeidentifiedcorrectlybyconventionaltechniques.ApplicationsAutomateddetectionofosteoarthritisfromknee-xray.Microcalcificationclustersinmammograms.EmphysemainCT-alungpathologycharacterisedbydestructionoflung.lungnoduledetectionfrompostero-anteriorchestradiographs.Cryo-electronmicroscopy:
instructuralbiologyisoneofthefieldswhichrequiresfullyautomatedobjectdetectiontechniques.SimulationResultsFig2:
Correlationgraphfortwosimilarimageslengthofthepeak=1Resultsforilluminationinvariancefig5:
(a)and(b)aretheX-rayimagesofthesamechestwithvariationinillumination.fig6:
Simulationresultfortheimagesinfig5ApplicationofPOCforpulmonaryemphysemadetection.Thereare80millionpatentswithdevelopedpulmonaryemphysemaallovertheworld3millionpatientsaredyingeveryyearStudieshaveshownthatlungtissueisaboutathirdofthelungvolumehastobedestructedbeforeemphysemacouldbedetected11.Doctorshavetwomethodstodiagnosepulmonaryemphysema.Oneofthemethodsisspirometry,anotherisdiagnosticimaging.Formerisaquantitativemethod,however,wecannotmakeanearlydetectionofthedisease.TheclassicalmethodofCTimageobjectiveevaluationisthePI(pixelindex)method.EmphysemashowsuponCTasareaswithlowattenuationcoefficients,withabnormaldistribution.Bydeterminingthenumberofpixelswithlowattenuation,emphysemacanbedetected.PIdeterminestheaveragenumberofpixelswithlowerattenuationthanthelimitvalue(lim).Fig.9:
SubjectA,normallung.Theseimagesshowhomogeneousdistributionofairinthelung.Thewhitedotsrepresentareaswithlowerattenuationvaluesthan950HU,930HU,and910HU,respectively.Thecalculatedpercentagesare:
3%,5%,and10%.13Fig.10:
SubjectC,severeemphysema.Thereisanobviousdestructionofthelungparenchyma.Pixelindexes:
34%,48%and62%.13IssueswithPImethodThePImethodisagood,well-knownmeasureofemphysema.Butitisnotabletodetectemphysemaincasesinwhichemphysemaandfibrosisoccuratthesametime.Thisisbecausefibrosistendstoincreasetheattenuationco-efficientofpixelswhereasemphysematendstoreducetheattenuationco-efficientofpixels.ToreducethehealthriskfromexposuretoradiationwhilemakingaCTscan,itisdesirabletousearadiationdosethatisaslowaspossible.However,theconstraintonirradiationdoseleadstoconsiderablenoiseinCTscans.RadiationdosagesdirectlyinfluencethePIindex.Fig.11:
Coronalsliceanditsaccompanyingemphysemamapcalculatedforathresholdof930HU.(a),(b)Scanwithclinicalradiationdose(PI=13.3%).(c),(d)Approximatelycorrespondingsliceofascanofthesamepatientwithatentimeslowerradiationdose(PI=15.8%)Fig.12:
(1a),(1b),(1c)SimulationresultsforemphysemadetectionFig.123:
Simulationresultsforemphysemaprogression.(4a)istheimageofahealthyindividual.(4b),(4c)and(4d)areimagesoftheemphysemaindifferentprogressivestages.(4e)showsthatthere11.7%differencebetween(4a)and(4b).(4f)showsthatthereis.58%differencebetween(4b)and(4c).(4g)showsthatthereis.02%changebetween(4c)and(4d)POCmethodgivesadirectmappingbetweenthenumberofaffectedpixelsandthepercentageofvisiblepixelsonthePOCmap.POCmethodscoresoverthetraditionalPImethodintermsoftheRadiationdosage.PImethodgivesbetterperformanceathigherdosagesofradiation.ButthePOCmethodisnotdependantonthebrightnessoftheimage.HenceitisabettermethodcomparedtothePImethod.Thistechniquecanbeextendedandverifiedoverotherpathologieslikeosteoarthritis,LungNoduleDetectionandMicrocalcificationclustersinmammograms.References1Fazl-e-Basit,M.Y.JavedandU.Qayyum,“FaceRecognitionUsingProcessedHistogramAndPhase-OnlyCorrelation(POC)”,InternationalConferenceonEmergingTechnologies,ICET2007,pp238-242,Nov.2007.2C.Nakajimaetal,“ObjectRecognitionAndDetectionByaCombinationOfSupportVectorMachineAndRotationInvariantPhaseOnlyCorrelation“.15thInternationalConferenceonPatternRecognition,IEEEProc,Vol.4,pp787-790,Sept.2000.3J.Z.Wangetal,InvestigationOfAPhase-OnlyCorrelationTechniqueForAnatomicalAlignmentOfPortalImagesInRadiationTherapy,Phys.Med.Biol.Vol.41,pp1045-1058,Jun.1996.4H.Nakajimaetal,“AFingerprintMatchingAlgorithmUsingPhaseOnlyCorrelation“.IEICETrans.Fundamentals,Vol.87-A,pp682-691,Mar.2004.5S.Watanabe,T.TanakaandE.Iwata,“BiometricAuthenticationUsingPhaseOnlyCorrelationWithCompensationAlgorithmForRotation.”SICE-ICASE
- 配套讲稿:
如PPT文件的首页显示word图标,表示该PPT已包含配套word讲稿。双击word图标可打开word文档。
- 特殊限制:
部分文档作品中含有的国旗、国徽等图片,仅作为作品整体效果示例展示,禁止商用。设计者仅对作品中独创性部分享有著作权。
- 关 键 词:
- 推广 营销 经验 分享
![提示](https://static.bdocx.com/images/bang_tan.gif)