基于ABVS影像组学列线图术前预测T1期乳腺癌腋窝淋巴结转移  

Preoperative prediction of axillary lymph node metastasis in T1 breast cancer based on ABVS imaging nomogram

作　　者：张宇[1] 汪珺莉 范莉芳 陈国仙 ZHANG Yu;WANG Junli;FAN Lifang;CHEN Guoxian(Department of Ultrasound Medicine,Wuhu Hospital Affiliated to East China Normal University(Wuhu Second People's Hospital),Anhui Wuhu 241000,China;Wannan Medical College,Anhui Wuhu 241002,China)

机构地区：[1]华东师范大学附属芜湖医院(芜湖市第二人民医院)超声医学科,安徽芜湖241000 [2]皖南医学院,安徽芜湖241002

出　　处：《现代肿瘤医学》2024年第16期3085-3092,共8页Journal of Modern Oncology

基　　金：安徽省高校重点科研项目(编号:2023AH051743);皖南医学院中青年科研基金项目(编号:WK202213);安徽省高等学校质量工程项目(编号:2022xsxx243)。

摘　　要：目的:探讨基于自动乳腺全容积成像(automatic breast volume scanner,ABVS)影像组学构建的列线图模型术前预测T1期乳腺癌腋窝淋巴结(axillary lymph node,ALN)转移的价值。方法:收集华东师范大学附属芜湖医院158例T1期乳腺癌患者的临床病理及影像资料,按7比3将患者随机分为训练组(n=110)及验证组(n=48)。利用MaZda纹理分析软件基于ABVS最大冠状面图像提取影像组学特征,并采用最小绝对收缩和选择算子算法(least absolute shrinkage and selection operator,LASSO)回归降维筛选最优特征,并构建影像组学标签评分(radiomics score,radscore)。通过单因素与多因素Logistic回归分析筛选预测T1期乳腺癌ALN转移的独立危险因子,构建联合预测模型,并绘制列线图。用受试者工作特征(receiver operating characteristic,ROC)曲线及曲线下面积(area under curve,AUC)进行模型性能评价,Hosmer-Lemeshow检验进行拟合优度评价。校准曲线评价模型的校准度,Delong检验比较模型的诊断效能,决策曲线分析(decision curve analysis,DCA)评估模型的临床实用性。结果:单因素与多因素Logistic回归发现象限、冠状面汇聚征、超声淋巴结阳性为独立危险因素。联合模型训练组、验证组AUC分别为0.944、0.862,Delong检验联合模型诊断效能最高,Hosmer-Lemeshow检验模型均拟合较好(训练组χ^(2)=6.877,P=0.550;验证组χ^(2)=13.904,P=0.084),校准曲线及DCA表明列线图具有较高的校准度及较好的临床适用性。结论:基于ABVS影像组学列线图术前能有效的预测T1期乳腺癌ALN转移风险,构建的列线图能够可视化预测结果,为精准诊疗提供无创手段。Objective:To evaluate the value of a nomogram model based on automatic breast volume scanner(ABVS)imaging for predicting axillary lymph node(ALN)metastasis in T1 breast cancer.Methods:Clinical pathological and imaging data of 158 patients with T1 breast cancer in Wuhu Hospital Affiliated to East China Normal University were collected.The patients were randomly divided into training group(n=110)and validation group(n=48)according to 7∶3.Using MaZda texture analysis software to extract radiomics features from ABVS maximum coronal images,and the least absolute shrinkage and selection operator(LASSO)regression dimension reduction algorithm was used to screen the best features.The radiomics label score(radscore)was constructed.The independent risk factors for predicting ALN metastasis in T1 breast cancer were screened by univariate and multivariate Logistic regression analysis,and the combined prediction model was constructed and a nomogram of the model was drawn.Receiver operating characteristic(ROC)curve and area under curve(AUC)were used to evaluate the diagnostic efficiency of the model.The goodness of fit of the model was evaluated by Hosmer-Lemeshow test.Calibration curve was used to evaluate the calibration degree of the model,the Delong test was used to compare the diagnostic efficiency of the model,and the decision curve analysis(DCA)was used to evaluate the clinical utility of the model.Results:Univariate and multivariate Logistic regression found that quadrant,coronal convergence sign,and ultrasound lymph node positivity were independent risk factors.The AUC of the combined model in training group and validation group were 0.944 and 0.862,respectively.Delong test combined model has the highest diagnostic efficiency.Hosmer-Lemeshow test model fits well(Training groupχ^(2)=6.877,P=0.550;Validation groupχ^(2)=13.904,P=0.084).The calibration curve and DCA indicate that the nomogram has high calibration accuracy and good clinical applicability.Conclusion:ABVS imaging nomogram can effectively predict the risk

关 键 词：乳腺癌 腋窝淋巴结 影像组学 ABVS 

分 类 号：R737.9[医药卫生—肿瘤]