乳腺x线图像微钙化点检测方法的研究

《乳腺x线图像微钙化点检测方法的研究》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、』E空窒堑厶：!；兰熊L生位造堡△BS!B△￡IABSTRACTABSTRACT：Breaslc3ncel-basbeconletbeleadingcaLiseol’deathamong、VOll]cnmtbecityMlicbisaseriousthrealtoIheph}sicalandmentalhealtholmodemv。onlellltisknownIllattheIlyingpatternanden'。irollmetltcOllllibutedtotheBreaslcanl2elinsome、v8*butitisimpossibletolindoUlthecxaclC

2、aLlScIheke>slbldecteasingthemolaalit、

3、areearlydscover3．eallydiagnosisandeall)tlcalmentBecauseofthelindtationoftadiologisls’Ⅲkexperienceandconsistentworkcompuler-aidedteclmolog>playsavery"importantroleintheearydiagnosisolmammogramsMicro—calcificationspresenta】1eallysignofbrcastcancer．Becauseofth

4、elowconlrastwithissuesandotherI．tctorsfindingalleffective'saytodetecltbemicro-calcificationsinmammogramsusingcomptReraidedtechnologiesisont2ofthemostsignificantobjectsontheresearcholbreastcancer．ThispaperisIbcusedontheimageprocessingandanal)sisofthemicro—calcificationsautodetectionbasedonthedif

5、liculliesofmicro-calcificationdetectionandthecharactmisticsanalysisolbiomedicalimagesandnaammogramsinthreemodulesFhemainresearchworkandlllnOVationsaleasfoIlows：ITkMammogramPreprocessingModuleBased011thepresentsituationthalmostextractionmethodsolbreaslregionarebighlycomplexthispaperpresentsanexl

6、ractionmethodofbreastregionbasedonmorphologyThcexperimema]resultsshowtheproposedmethodhasobtainedthehighlyperlbrmanceonthespeedolextraclionandtheuniversalitybasedOllthecompleteandefl)ctiveextractionofbreastregion2IbeRegionol’InterestExt；actionModuleBasedonthehctthatmedica】imagesusuallyCOtlSiSto

7、lweightedindependenlcomponentsandIbestrengthofindependentconlponcntanabsisonbighdataslatisticalcorrelationthispaperproposedanletbodtoextFacttheregionolinlc]cstbasedonindependentcomponentanalysisUsing11asllCAbasedoninaximizalionol'negativeetropytoensurethespeedofICAconvt：lgenceWeseparatedtheIhls

8、esignalsby,meansolartificialneura]networksandsupportt,ectolmachineclassifiersinordertoextractoltheregionofinterestTheexperhnenta[resultssbowthatsuppottveCtOlnlaebbaeclassifierdoesabetterperlbrmancedlanartificialneuralnetworksinbol