基于雙視圖的乳腺腫塊檢測方法研究.pdf

1、華中科技大學(xué)碩士學(xué)位論文基于雙視圖的乳腺腫塊檢測方法研究姓名：文耀光申請學(xué)位級別：碩士專業(yè)：計(jì)算機(jī)應(yīng)用技術(shù)指導(dǎo)教師：許向陽20110117華中科技大學(xué)碩士學(xué)位論文IIAbstractBreastcancerisoneofthemostcommoncancersfwomen.Mostbreastcancerscanbedetectedfrombreastmasses.Mammogramsareaneffectivemethodofearl

2、ydetectionofbreastcancer.Radiologistsoftencomparethetwoviewsofthesamebreastwhenreadingmammograms.DualviewbasedCADwasdevelopedfromthatmodel.ThefirststepofourdualviewbasedCADwasdetectingcidatemassesseparatelyfromthetwoview

3、softhesamebreast.Aregionrestrictedsegmentationmethodwasusedtosegmentthebreastregion.Thensuspiciousmasseswerelocatedbyamultiscaletemplatematchingmethod.Logisticclassifierwasusedtocalculatethedetectionsceofeachsuspiciousma

4、ss.Accdingtothefactthatthesamemassindifferentviewshasidenticaldistancefromthenipplesuspiciousmassesdetectedfmdualviewswerematchedasmatchedpairs.FeatureswereextractedfromeachmatchedpairthenclassifiedbyLogisticclassifier.F

5、alsepositiveswerereducedbythedetectionscesofsuspiciousmassmatchedpairs.184caseswerechosenfromDigitalDatabasefScreeningMammographyofSouthFlidaUniversitywhenthesensitivityofsingleviewbasedCADreached90%average2.9falsepositi

6、veregionsweredetectedineachimage.FthedualviewbasedCADthesensitivityfalsepositivewere90%1.98respectively.AverageonefalsepositiveisreducedindualviewbasedCADasinsingleviewbasedCAD.Keywds:BreastcancerMammogramMassclassificat