基于多模态超声鉴别乳腺腺病与小乳腺癌的列线图模型构建及应用  

Construction and Application of a Column-Line Diagram Model Based on Multimodal Ultrasound to Discriminate Breast Adenopathy from Small Breast Cancer

作　　者：王孝荣 王佳佳 詹韵韵[1] 毕玉[1] 彭梅[1] Wang Xiaorong;Wang Jiajia;Zhan Yunyun;Bi Yu;Peng Mei(Department of Ultrasound Medicine,The Second Affiliated Hospital of Anhui Medical University,Hefei 230601,China)

机构地区：[1]安徽医科大学第二附属医院超声医学科,合肥市230601

出　　处：《中国超声医学杂志》2024年第9期993-997,共5页Chinese Journal of Ultrasound in Medicine

基　　金：安徽省2023年临床医学研究转化专项(No.202304295107020016);安徽医科大学校级质量工程项目(No.2024xjxm46);安徽医科大学第二附属医院临床研究培育计划(No.2021LCZD06);2023年度安徽省高校自然科学研究项目(No.2023AH040373)。

摘　　要：目的探讨基于多模态超声构建的列线图预测模型在鉴别乳腺腺病与小乳腺癌上的价值。方法筛选经病理证实为乳腺腺病或小乳腺癌的患者212例(214个乳腺肿块),分析其超声图像特征,并联合血管指数,按照7∶3比例随机分为训练集(n=145)及测试集(n=69),对训练集进行单因素及多因素Logistic分析,筛选独立影响因素并构建列线图,绘制该模型的ROC曲线和校准曲线,并利用测试集进行内部验证。结果患者年龄及肿块微钙化、血管指数(VI)和弹性评分是鉴别乳腺腺病和小乳腺癌的独立预测因素。基于上述变量构建的列线图预测模型训练集AUC为0.948(95%CI:0.914~0.981),C-index为0.948,校准曲线显示模型预测概率与实际概率之间具有良好的一致性;测试集AUC为0.874(95%CI:0.790~0.957),C-index为0.874,内部验证显示列线图具有较高的诊断价值。结论基于多模态超声构建的列线图预测模型可以有效鉴别乳腺腺病和小乳腺癌,有利于患者的个体化治疗,具有一定的临床应用价值。Objective To investigate the value of the prediction model based on multimodal ultrasound in distin-guishing breast adenopathy from small breast cancers.Methods A total of 212 patients(214 breast masses)with pathologically confirmed breast adenopathy or small breast cancers were screened.Their ultrasound image characteris-tics and the vascular index were analyzed.They were randomly divided into a training set(n=145)and a test set(n=69)according to the ratio of 7∶3.Single-factor and multifactorial Logistic analyses were performed on the training set to identify the independent influencing factors and construct a column-line diagram.ROC curves and calibration curves was drawn for the model,and the test set data was used to perform internal validation using.Results Patient age,microcalcifications,vascular index(Ⅵ)and elasticity score were independent predictors to discriminate between breast adenopathy and small breast cancer.The training set AUC of the column-line diagram prediction model constructed based on the above variables was 0.948(95%CI:0.914~0.981),and the C-index was 0.948,and the calibration curve showed good agreement between the model predicted probability and the actual probability;The AUC of the test set was 0.874(95%CI:0.790~0.957)and the C-index was 0.874,and the internal validation showed that the col-umn-line diagram had a high diagnostic value.Conclusions The prediction model of column-line diagram constructed based on multimodal ultrasound can effectively identify breast adenopathy and small breast cancer,which is conducive to the individualized treatment of patients and has certain clinical application value.

关 键 词：超声特征 乳腺腺病 小乳腺癌 血管指数 列线图 

分 类 号：R445.1[医药卫生—影像医学与核医学] R737.9[医药卫生—诊断学]