乳腺MRI成簇环状非肿块强化病变的恶性风险预测模型构建与评估  

Construction and Evaluation of Malignant Risk Prediction Model of Breast MRI Clustered Ring Non-mass Enhancement Lesions

作　　者：黄瑞岁 陆善金 慕鉴 丁可 赵沁萍 陈江 HUANG Ruisui;LU Shanjin;MU Jian(Department of Radiology,the Third Affiliated Hospital of Guangxi Medical University,Nanning 530031,P.R.China)

机构地区：[1]广西医科大学第三附属医院放射科,南宁530031

出　　处：《临床放射学杂志》2024年第3期341-345,共5页Journal of Clinical Radiology

摘　　要：目的 构建乳腺成簇环状非肿块强化病变的恶性风险预测模型,并评估该模型的预测效果。方法 回顾性分析行MRI检查并经手术或活检病理证实的良、恶性乳腺成簇环状非肿块强化病灶共107个,采用多因素Logistic回归分析筛选恶性病变的危险因素,采用R软件构建预测恶性病变的列线图模型,分别使用受试者工作特征曲线、校准曲线、决策曲线分析评估模型的区分度、校准度、临床实用性。结果 单因素分析显示,最大径、ADC值、早期强化率及总体分布特征、时间信号强度曲线(TIC)类型差异具有统计学意义(P<0.05)。多因素Logistic回归分析显示,ADC值、最大径、TIC类型是成簇环状非肿块强化恶性病变的独立危险因素(P<0.05)。列线图模型预测评估的区分度曲线下面积为0.911,对应的敏感度、特异度分别为89.10%、85.20%;校准曲线预测值与实际值基本一致,Hosmer-Lemeshow拟合优度检验χ~2=13.27,P>0.05;与各独立危险因素比较,模型的决策曲线离两种极端情况最远,在阈值0~80%区间内体现出更高的净获益。结论 ADC值、最大径、TIC类型是乳腺成簇环状非肿块强化恶性病变的独立危险因素,由此构建的恶性风险预测列线图模型具有较好的区分度、校准度及临床实用性,可为临床制定个体化的诊治方案提供重要参考。Objective To establish a malignant risk prediction model of breast clustered ring non-mass enhancement lesions,and to evaluate the predictive effect of the model.Methods A retrospective analysis was performed in 107 benign and malignant breast clustered ring non-enhancement lesions confirmed by operation or biopsy pathology by MRI.Multivariate Logistic regression analysis was used to screen the risk factors of malignant lesions.R software was used to construct a nomogram model for predicting malignant lesions.The receiver operating characteristic curve,calibration curve and decision curve analysis were used to evaluate the differentiation,calibration and clinical practicability of the model.Results Single factor analysis showed that there were significant differences in maximum diameter,ADC value,early-enhancement-rate,distribution characteristics and TIC types(P<0.05).Multivariate Logistic regression analysis showed that ADC value,maximum diameter and TIC type were independent risk factors for clustered ring non-mass enhancement malignant lesions(P<0.05).The area under the receiver operating characteristic curve of the predicting and evaluating discriminative degree in nomogram model was 0.911,and the corresponding sensitivity and specificity were 89.10% and 85.20% respectively.The predicted value of the calibration curve was basically consistent with the actual value,and the Hosmer-Lemeshow goodness-of-fit test showed =13.27,P>0.05.Compared with the independent risk factors,the decision curve of the model was farthest from the two extremes,showing a higher net benefit in the threshold range of 0-80%.Conclusion ADC value,maximum diameter and TIC type are independent risk factors for breast clustered ring non-mass enhancement malignant lesions.The malignant risk prediction nomogram model has good differentiation,calibration and clinical practicability,and can provide important reference for the clinical formulation of individual diagnosis and treatment plan.

关 键 词：乳腺 磁共振成像 成簇环状强化 列线图 

分 类 号：R737.9[医药卫生—肿瘤] R445.2[医药卫生—临床医学]