人工智能论文英文版-FPDANet：A Multi-Section Classification Model for Intelligent Screening of Fetal Ultrasound.pdfVIP

FPDANet:AMulti-SectionClassificationModelfor

IntelligentScreeningofFetalUltrasound

∗†∗‡‡§¶

MinglangChen,JieHe,CaixuXu,BochengLiang,ShengliLi,GuannanHe,andXiongjieTao

∗(a)GuangxiKeyLaboratoryofMachineVisionandIntelligentControl,WuzhouUniversity,Wuzhou,543002China

†(b)CollegeofComputerScienceandElectronicEngineering,HunanUniversity,Changsha,410082China

‡(c)DepartmentofUltrasound,ShenzhenMaternalandChildHealthcareHospital,Shenzhen,518100China

§(d)DepartmentofUltrasound,SichuanProvincialMaternityandChildHealthcareHospital,Chengdu,610100China

¶(e)FacultyofHumanitiesandArts,MacauUniversityofScienceandTechnology,Macau,999078China

5

2Abstract—ResNethasbeenwidelyusedinimageclassification[6].Thismakesittime-consumingandlaboriousfordoctors

0tasksduetoitsabilitytomodeltheresidualdependenceoftoselectthesections[1],andthecorrectnessandauthorityof

2constantmappingsforlinearcomputation.However,theResNetsectionselectiondependsonthedoctor’sclinicalexperience

methodadoptsaunidirectionaltransferoffeaturesandlacks

naneffectivemethodtocorrelatecontextualinformation,whichandprofessionalism,whichinturnincreasesthedifficultyof

uisnoteffectiveinclassifyingfetalultrasoundimagesinthefetalprenatalscreening.Deeplearningmethodsareutilizedto

J

classificationtask,andfetalultrasoundimageshaveproblemsautomaticallyandrapidlyselectanddifferentiateultraso