·85CHINESEJOURNALOFCTANDMRI,FEB.2023,Vol.21,No.02TotalNo.160【第一作者】蔡振德，男，主治医师，主要研究方向：乳腺影像诊断。E-mail：endece@163.com【通讯作者】罗慧，女，主任医师，主要研究方向：乳腺及浅表器官超声诊断。E-mail：luoh98@sina.com论著MultimodalImagingStudyofTripleNegativeBreastCarcinomaBasedonDeepLearningTechnologyCAIZhen-de1,2,MAJie2,LUOHui3,*.1.ShantouUniversityMedicalCollege,Shantou515000,GuangdongProvince,China2.DepartmentofRadiology,ShenzhenPeople'sHospital,Shenzhen518020,GuangdongProvince,China3.DepartmentofUltrasound,ShenzhenPeople'sHospital,Shenzhen518020,GuangdongProvince,ChinaABSTRACTObjectiveTocomparethediagnosticefficacyofdigitalmammography(DM)combinedwithartificialintelligencediagnosticsystem(AI)withultrasound(US)anddynamicenhancedMRI(DCE-MRI)inthediagnosisoftriple-negativebreastneoplasms(TNBC),andevaluatetheroleofAI.Methods42casesofTNBCand68casesofbenignbreastcancerwerecollected.AllcaseshadcompletepreoperativedataofDM,USandDCE-MRI.ThedataofDM,USandDCE-MRIofbreastwerereviewed,andasetofdiagnosticresultswereobtainedrespectively.Inaddition,theDMimageswerediagnosedbyAIandAIjointradiologistrespectively,andtwosetsofresultswereobtained.Thesubjectoperatingcharacteristiccurves,areaundercurve(AUC),specificity,sensitivity,positivepredictivevalueandnegativepredictivevalueofthefivediagnosticmethodswerecomparedandanalyzed,andtheconsistencyofeachdiagnosticmethodwascompared.ResultsTheAUCvalueofDM(radiologist)groupwasthelowest,whiletheAUCvalueofDM(AI+radiologist)groupwasthehighest,whichwas0.770,betterthanUSandDCE-MRI.ThesensitivityofDCE-MRIandUSexaminationwasthehighest,butitsspecificitywaspoor.ConclusionInthestudy,DM(radiologist)combinedwithAIcaneffectivelyimproveitsdiagnosticefficacyintriple-negativebreastcancer,anditscomprehensivediagnosticefficacyisslightlybetterthanthatofMRI.Keywords:TripleNegativeBreastNeoplasms;Ultrasound;Mammography;MRI;ArtificialIntelligence乳腺癌是我国女性中发病率最高的恶性肿瘤[1]。三阴性乳腺癌(triple-negativebreastcarcinoma,TNBC)是...
