موضوع فارسی :تشخیص توده در تصویر ماموگرافی با استفاده از سی ان ان، geostatistic
توابع و SVM
موضوع انگلیسی :Detection of massesin mammogram image susing CNN,geostatistic
functions andSVM
تعداد صفحه :12
فرمت فایل :PDF
سال انتشار :2011
زبان مقاله : انگلیسی
Breastcanceroccurswithhighfrequencyamongtheworld’spopulationanditseffectsimpactthe
patients’perceptionoftheirownsexualityandtheirverypersonalimage.Thisworkpresentsa
computationalmethodologythathelpsspecialistsdetectbreastmassesinmammogramimages.The
first stageofthemethodologyaimstoimprovethemammogramimage.Thisstageconsistsinremoving
objects outsidethebreast,reducingnoiseandhighlightingtheinternalstructuresofthebreast.Next,
cellularneuralnetworksareusedtosegmenttheregionsthatmightcontainmasses.Theseregionshave
their shapesanalyzedthroughshapedescriptors(eccentricity,circularity,density,circulardispropor-
tion andcirculardensity)andtheirtexturesanalyzedthroughgeostatisticfunctions(Ripley’s K function
and Moran’sandGeary’sindexes).Supportvectormachinesareusedtoclassifythecandidateregionsas
massesornon-masses,withsensitivityof80%,ratesof0.84falsepositivesperimageand0.2false
negativesperimage,andanareaundertheROCcurveof0.87
