《中国医学创新》第21卷第24期(总第666期)2024年8月经验交流Jingyanjiaoliu
乳腺AI与钼靶在乳腺结节定性诊断及术前
指导中的应用价值
①①①
李琪李丰章刘天笑
【摘要】目的:
观察乳腺人工智能(AI)与钼靶在乳腺结节定性诊断及术前指导中的应用价值。
方法:选取2019年1月—2023年6月萍乡市人民医院收治的乳腺结节患者164例,对所有患者行乳腺
AI检查及乳腺钼靶检查,以病理诊断结果为金标准,比较乳腺AI、钼靶及两者联合对乳腺结节定性诊断
结果:
及术前指导的价值。164例患者中,经手术病理结果发现174个乳腺结节,共136个良性病变(纤
维腺瘤54例,乳腺病26例,肉芽肿性小叶性乳腺炎52例,单纯性囊肿4例),38个恶性病变(浸润性
导管癌14例,乳头状癌12例,导管原位癌12例)。乳腺AI检查定性诊断乳腺结节的敏感度、特异度及
准确性分别为73.68%、91.17%及87.36%;乳腺钼靶检查为57.89%、85.29%及79.31%;两者联合敏感
度及准确性分别为89.47%、91.95%,均显著高于单独诊断,差异均有统计学意义(P0.05)。结论:乳
腺AI联合钼靶检测能显著提高乳腺结节定性诊断效能,对于乳腺结节术前评估价值较高。
【关键词】
乳腺AI乳腺钼靶乳腺结节定性诊断术前指导
TheApplicationValueofBreastAIandMolybdenumTargetinQualitativeDiagnosisand
PreoperativeGuidanceofBreastNodules/LIQi,LIFengzhang,LIUTianxiao.//MedicalInnovationof
China,2024,21(24):139-142
[Abstract]Objective:Toobservetheapplicationvalueofbreastartificialintelligence(AI)and
mammographyinqualitativediagnosisandpreoperativeguidanceofbreastnodules.Method:Atotalof164
patientswithbreastnodulesadmittedtoPingxiangPeoplesHospitalfromJanuary2019toJune2023wereselected,
andallpatientsunderwentbreastAIandmammography.Thepathologicaldiagnosisresultswereusedasthegold
standard.ThevalueofbreastAI,mammographyandtheircombinationinqualitativediagnosisandpreoperative
guidanceofbreastnoduleswerecompared.Result:Among164patients,174breastnoduleswerefoundthrough
surgicalpathology,withatotalof136benignlesions(54fibroadenomas,26breastdiseases,52granulomatous
lobularmastitisand4simplecysts)and38malignantlesions(14invasiveductalcarcinoma,12papillarycarcinoma
and12ductalcarcinomainsitu).Thesensitivity,specificityandaccuracyofqu