资源描述:
《Cancer Prediction of breast cancer malignancy using an artificial neural network》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、2944PredictionofBreastCancerMalignancyUsinganArtificialNeuralNetworkCareyE.Floyd,Jr.,Ph.D.,*,tJosephY.Lo,Ph.D.,*,tA.JoonYun,M.D.,*DanielC.Sullivan,M.D.,*andPhyllisJ.Kornguth,M.D.,Ph.D.*Background.Anartificialneuralnetwork(ANN)wastechniqueforearlydetectionofbreastcancer,asignifi-developedtopred
2、ictbreastcancerfrommammographiccantfractionofpatientsreferredforbiopsyastheresultfindings.Thisnetworkwasevaluatedinaretrospectiveofmammographyfindingsdonothaveamalignancy.study.Althoughspecific,biopsyisaninvasive,costly,andMethods.Forasetofpatientswhowerescheduledemotionallystressfulprocedure.
3、Inanefforttoreduceforbiopsy,radiologistsinterpretedthemammogramsandthenumberofbenigncasessenttobiopsy,anartificialprovideddataoneightmammographicfindingsaspartintelligencetechniquewasinvestigatedtopredicttheofthestandardmammographicworkup.Thesefindingsoutcomeofbiopsyfromradiographicfindings.An
4、arti-wereencodedasfeaturesforanANN.Resultsofbiopsiesweretakenastruthinthediagnosisofmalignancy.Theficialneurafnetwork(ANN)wasdevelopedthattakesANNwastrainedandevaluatedusingajackknifesam-radiologicfindingsasinputsandpredictstheoutcomeplingonasetof260patientrecords.Performanceoftheofbiopsyasano
5、utput.networkwasevaluatedintermsofsensitivityandspeci-ANNsarecomputeralgorithmswhosestructureficityoverarangeofdecisionthresholdsandwasex-andfunctionarebasedonmodelsofthestructureandpressedasareceiveroperatingcharacteristiccurve.learningbehaviorofbiologicalneuralnetworks.TheseResults.TheANNper
6、formedmoreaccuratelythanalgorithmsaretypicallyemployedtoclassifyasetoftheradiologists(P<0.08)witharelativesensitivityof1.0patternsintooneofseveralclasses.Theclassificationandspecificityof0.59.rulesarenotwrittenintothealgorithm,butarelearnedConclusions.AnANNcanbetrainedtopredictma-bythenetworkf
7、romexamples,lignancyfrommammographicfindingswithahighde-Artificialneuralnetworkshavebeendevelopedforgreeofaccuracy.Cancer1994;74:2944-8.awidevarietyofcomputationalproblemsincognition,Keywords:breastneoplasms,diagnosis;computers,neu-patt