基于增强CT影像组学的列线图模型在预测胸腺上皮性肿瘤WHO简化分型中的应用研究  

An Application Study of Nomogram Model Based on Enhanced CT Radiomics in Predicting WHO Simplified Typing of Thymic Epithelial Tumors

作　　者：郑璇 徐加利[1] 刘浩 谢宗玉[1] ZHENG Xuan;XU Jia-li;LIU Hao;XIE Zong-yu(Department of Radiology,The First Affiliated Hospital of Bengbu Medical College,Bengbu 233099,Anhui Province,China;Beijing Yizhun-AI Technology Limited Company,Beijing 10089,China)

机构地区：[1]蚌埠医学院第一附属医院放射科,安徽蚌埠233099 [2]北京医准智能科技有限公司,北京10089

出　　处：《中国CT和MRI杂志》2023年第10期102-105,共4页Chinese Journal of CT and MRI

基　　金：2021年蚌埠医学院自然科学重点项目(2021byzd093)。

摘　　要：目的 探讨基于增强CT图像的列线图模型在预测胸腺上皮性肿瘤(TETs)WHO简化分型中的应用价值。方法 回顾性分析术前行胸部增强CT检查并经手术病理证实的165例TETs,按照8:2的比例随机划分为训练集132例与验证集33例。于静脉期图像手动勾画感兴趣区(ROI)并提取影像组学特征,经数据降维筛选出有效特征并建立影像组学公式,计算每位患者的得分(radscore)并构建影像组学模型。纳入多个CT特征,经多因素逻辑回归筛选出具有独立预测价值的特征并构建CT特征模型。联合具有独立预测价值的CT特征及radscore构建列线图模型。采用受试者工作特性曲线(ROC)评估模型的诊断效能,校正曲线及决策曲线(DCA)评估模型的预测准确性及临床应用价值。结果 在训练集中,纵隔脂肪浸润与radscore共同构建了列线图模型,模型在训练集的曲线下面积(AUC)值为0.902(95%CI:0.838~0.947),在验证集的AUC值为0.824(95%CI:0.652~0.934)。结论 基于增强CT影像组学的列线图模型对于TETs WHO简化分型有较高的预测价值,可作为一种无创的术前评估工具辅助临床决策制订。Objective To explore the application value of nomogram model based on enhanced CT radiomics in predicting WHO simplified typing of thymic epithelial tumors(TETs).Methods We retrospectively analyzed 165 cases of TETs that underwent preoperative enhanced chest CT and pathologically confirmed.165 cases were randomly divided into training set(n=132)and validation set(n=33)according to the ratio of 8:2.Region of interest(ROI)was manually segmented in venous phase images,and radiomics features were extracted.Effective radiomics features were screened by dimension reduction,and radscore formula was formed.The score of each patient was calculated and the radiomics model was established.Multiple CT findings were included and analyzed.The CT findings with independent predictive value were selected by multivariate logistic regression,and the CT findings model was built.The selected CT findings and radscore constructed nomogram model.The diagnostic performance of models evaluated from Receiver operating characteristic curve(ROC),calibration curve and decision curve(DCA)assess the prediction accuracy and clinical value of application.Results In training set,two factors,infiltration and radscore were used to construct nomogram model.In training set,the area under curve(AUC)value of the nomogram model was 0.902(95%CI:0.838~0.947),and validation set’s AUC value was 0.824(95%CI:0.652~0.934).Conclusion The nomogram model based on enhanced CT radiomics has high predictive value for the WHO simplified typing of TETs,and it will be a non-invasive evaluation tool to assist clinical decision making before surgery.

关 键 词：影像组学 胸腺上皮性肿瘤 增强CT 列线图 

分 类 号：R736.3[医药卫生—肿瘤]