基于CT影像组学列线图预测早期非小细胞肺癌血管生成拟态表达  

Prediction of Vasculogenic Mimicry Expression in Early-Stage Non-Small Cell Lung Cancer Based on CT Radiomics Nomogram

作　　者：张群芳 徐鹤 周辉 刘德顺 张雪丽 谢宗玉 ZHANG Qunfang;XU He;ZHOU Hui(Department of Radiology,The First Affiliated Hospital of Bengbu Medical University,Bengbu,Anhui Province 233004,P.R.China)

机构地区：[1]蚌埠医科大学第一附属医院放射科,233004 [2]滁州城市职业学院附属凤阳县人民医院医学影像科,233100 [3]安徽理工大学第一附属医院影像中心,淮南232000

出　　处：《临床放射学杂志》2025年第4期644-650,共7页Journal of Clinical Radiology

基　　金：安徽省重点研究与开发计划项目(编号:2022e07020033);滁州市科技计划项目(编号:2022ZD007);安徽理工大学医学专项培育项目(编号:YZ2023H2C018)。

摘　　要：目的探讨CT影像组学列线图预测早期非小细胞肺癌(NSCLC)血管生成拟态(VM)表达的应用价值。方法纳入159例NSCLC患者,按照7∶3随机分为训练组(n=111)与验证组(n=48)。以训练组为研究队列,筛选与VM表达有关的独立临床预测因素及影像组学特征,分别构建临床模型、影像组学模型以及联合列线图模型。采用受试者工作特征(ROC)曲线、校准曲线及决策曲线对3种模型性能进行评价。结果临床模型由病灶最大径、毛刺征、分叶征及密度构成;影像组学模型由影像组学评分(Radscore)构成;列线图模型由Radscore、最大径、毛刺征以及分叶征构成。训练组中,列线图模型、影像组学模型以及临床模型的曲线下面积(AUC)分别为0.945、0.913以及0.851。验证组中,3种模型的AUC分别为0.964、0.873和0.855。校准曲线分析显示,列线图模型的预测结果和实际观察结果间具有良好的一致性,决策曲线显示该模型具有较高的临床获益。结论影像组学结合临床危险因素可以早期有效地预测NSCLC患者VM的表达状态,以辅助临床制定个性化治疗方案。Objective To investigate the application value of CT radiomics nomogram to predict vasculogenic mimicry(VM)expression in early stage non-small cell lung cancer(NSCLC).Methods 159 cases of NSCLC were included and randomized into training group(n=111)and validation group(n=48)according to 7∶3.With the training group as the study cohort,clinical independent predictors of VM expression as well as radiomics features were screened,and clinical model,radiomics model,and CT-radiomics nomogram model were constructed respectively.The performance of the three models was evaluated using the receiver operating characteristic(ROC)curve,calibration curves and decision analysis.Results The clinical model consisted of maximum diameter,the burr sign,the lobulation sign,and the density,the radiomics model consisted of radiomics score(Radscore),and the nomogram model consisted of Radscore,the maximum length,the burr sign,and the lobulation sign.In the training group,the area under the curve(AUC)of the nomogram model,radiomics model and clinical model were 0.945,0.913 and 0.851,respectively.In the validation group,the AUCs of the three models were 0.964,0.873 and 0.855,respectively.The calibration curve analysis showed that there was good agreement between nomogram model predictions and actual observations,and the decision curve analysis demonstrated high clinical benefit of the model.Conclusion Radiomics combined with clinical risk factors can effectively predict the expression status of VM in NSCLC patients at an early stage to assist the clinical development of personalized treatment plans.

关 键 词：影像组学列线图 非小细胞肺癌 血管生成拟态表达 

分 类 号：R73[医药卫生—肿瘤]