不同年资医师独立操作BI-RADS联合S-Detect诊断乳腺肿物比较OA
Comparison of BI-RADS combined with S-Detect for diagnosing breast masses by ultrasound physicians with different seniorities under independent operation
目的:比较不同年资超声医师在独立操作下应用乳腺影像报告和数据系统(BI-RADS)联合S-Detect诊断乳腺肿物的效能差异,评估该技术对缩小医师经验差距的价值.方法:前瞻性纳入2021年6月至2023年1月于汕头大学医学院第一附属医院拟行手术治疗的79例女性患者共92个乳腺实性肿物.由低年资(2年)、中年资(5年)、高年资(≥10年)超声医师分别独立进行灰阶超声扫查与S-Detect分析.以病理诊断为金标准,采用受试者工作特征曲线(ROC)及净重新分类改善指标(NRI)评估诊断效能,应用Kappa系数评价诊断一致性.结果:病理证实良性59个(64.1％)、恶性33个(35.9％).3位医师S-Detect结果整体一致性极高(Fleiss Kappa=0.822).联合S-Detect后,低年资医师的ROC曲线下面积(AUC)由0.733升至0.859(校正后P<0.05),NRI为37.2％;中年资医师AUC由0.852升至0.876(校正后P>0.05),NRI为6.4％;高年资医师AUC由0.961降至0.911(校正后P>0.05),NRI为-19.3％(P=0.008).低年资与高年资医师AUC差值从0.228缩小至0.052,诊断一致性由中等(Kappa=0.555)提升至高度一致(Kappa=0.777).结论:BI-RADS联合S-Detect可提升低年资医师的诊断准确性,有效缩小不同年资医师间的诊断差距,促进诊断水平均质化;但对高年资医师增益有限,可能引入误判,临床应分层应用.
Objective:To compare the diagnostic efficacy of the Breast Imaging Reporting and Data System(BI-RADS)combined with S-Detect for breast masses by ultrasound physicians with different seniorities under independent operation,and to evaluate the value of this technology in narrowing the experience gap among physicians.Methods:A total of 92 solid breast masses from 79 female patients who were scheduled for surgical treatment at the First Affiliated Hospital of Shantou University Medical College from June 2021 to January 2023 were prospectively enrolled.Ultrasonographers with low(2 years),intermediate(5 years),and high(≥10 years)seniority independently performed gray-scale ultrasound scanning and S-Detect analysis.Using pathological diagnosis as the gold standard,the diagnostic efficacy was evaluated by receiver operating characteristic(ROC)curve and net reclassification improvement(NRI),and the diagnostic consistency was assessed by Kappa coefficients.Results:Pathology confirmed 59 benign(64.1％)and 33 malignant(35.9％)masses.The overall consistency of S-Detect results among the three physicians was excellent(Fleiss Kappa=0.822).After combining with S-Detect,the area under the ROC curve(AUC)for low-seniority physicians significantly increased from 0.733 to 0.859(adjusted P<0.05),with an NRI of 37.2％;for intermediate-seniority physicians,the AUC increased from 0.852 to 0.876(adjusted P>0.05),with an NRI of 6.4％;for high-seniority physicians,the AUC decreased from 0.961 to 0.911(adjusted P>0.05),with an NRI of-19.3％(P=0.008).The AUC difference between low-and high-seniority physicians narrowed from 0.228 to 0.052,and the diagnostic consistency improved from moderate(Kappa=0.555)to substantial(Kappa=0.777).Conclusion:BI-RADS combined with S-Detect can improve the diagnostic accuracy of low-seniority physicians and effectively narrow the diagnostic gap among physicians with different seniorities,promoting homogenization of diagnostic performance.However,its added value for high-seniority physicians is limited and may introduce misjudgments;thus,it should be applied in a stratified manner in clinical practice.
林立坤;廖钰明;祁奇;练燕珊;黄喜锋;王金宏
汕头大学医学院第一附属医院超声科,广东 汕头 515041汕头大学医学院第一附属医院超声科,广东 汕头 515041汕头大学医学院第一附属医院超声科,广东 汕头 515041汕头大学医学院第一附属医院超声科,广东 汕头 515041汕头大学医学院第一附属医院超声科,广东 汕头 515041汕头大学医学院第一附属医院超声科,广东 汕头 515041
医药卫生
乳腺肿物超声检查乳腺影像报告和数据系统S-Detect人工智能诊断效能
breast massesultrasonographyBreast Imaging Reporting and Data SystemS-Detectartificial intelligencediagnostic efficacy
《汕头大学医学院学报》 2026 (2)
77-83,7
广东省医学科研基金(A2024154)
评论