肿瘤防治研究2023年第50卷第1期CancerResPrevTreat,2023,Vol.50,No.1·98·doi:10.3971/j.issn.1000-8578.2023.22.0704深度学习在膀胱癌病理学中的研究进展郑庆源，杨瑞，王磊，陈志远，刘修恒ResearchProgressofDeepLearninginBladderCancerPathologyZHENGQingyuan,YANGRui,WANGLei,CHENZhiyuan,LIUXiuhengDepartmentofUrology,RenminHospitalofWuhanUniversity,Wuhan430060,ChinaCorrespondingAuthor:LIUXiuheng,E-mail:drliuxh@hotmail.comAbstract:Theincidenceofbladdercancerisincreasingannually,andthegoldstandardforitsdiagnosisreliesonhistopathologicalbiopsy.Whole-slidedigitizationtechnologycanproducethousandsofhigh-resolutioncapturedpathologicalimagesandhasgreatlypromotedthedevelopmentofdigitalpathology.Deeplearning,asanewmethodofartificialintelligence,hasachievedremarkableresultsintheanalysisofpathologicalimagesfortumordiagnosis,moleculartyping,andpredictionofprognosisandrecurrenceofbladdercancer.Traditionalpathologyreliesheavilyontheprofessionallevelandexperienceofpathologists;assuch,itishighlysubjectiveandhaspoorreproducibility.Deeplearningcanautomaticallyextractimagefeatures.Itcanalsoimprovediagnosticefficiencyandrepeatabilityandreducemissedandmisdiagnosedrateswhenusedtoassistpathologistsinmakingdecisions.Thistechnologycannotonlyalleviatethepressureofthecurrentshortageofskilledworkforceandunevenmedicalresourcesbutalsopromotethedevelopmentofprecisionmedicine.Thisarticlereviewsthelatestresearchprogressandprospectsofdeeplearninginpathologicalimageanalysisofbladdercancer.Keywords:Artificialintelligence;Deeplearning;Bladdercancer;Pathologicalimages;Digitalpathology;PrecisionmedicineFunding:HubeiProvinceKeyResearchandDevelopmentProject(No.2020BCB051);HubeiProvinceCentralGuidanceLocalScienceandTechnologyDevelopmentSpecialProject(No.ZYYD2022000181);TheNationalMedicalEducationDevelopmentCenterMedicalSimulationEducationResearchProject(No.2021MNYB11)Competinginterests:Theauthorsdeclarethattheyhavenocompetinginterests.摘要：膀胱癌的发病率逐年上升，其诊断的金标准依赖于组织病理活检。全载玻片数字化技术可...
