CT影像组学联合临床特征模型可术前预测胸腺上皮性肿瘤的Ki-67表达水平  

Preoperative prediction of Ki-67 expression in thymic epithelial tumors by CT imaging histology combined with clinical features model

作　　者：吴春红 胡丛阳 许晓燕 胡雪超 周永 WU Chunhong;HU Congyang;XU Xiaoyan;HU Xuechao;ZHOU Yong(Imaging Diagnosis Center,The Third Clinical Medical College of Xinjiang Medical University(Affiliated Cancer Hospital of Xinjiang Medical University),Urumqi 830011,China)

机构地区：[1]新疆医科大学第三临床医学院(新疆医科大学附属肿瘤医院)影像诊断中心,新疆乌鲁木齐830011

出　　处：《分子影像学杂志》2025年第2期152-158,共7页Journal of Molecular Imaging

基　　金：新疆维吾尔自治区自然科学基金(2021D01C392);新疆维吾尔自治区产学合作协同育人项目(2023210016)。

摘　　要：目的探讨CT影像组学联合临床特征模型术前预测胸腺上皮性肿瘤(TET)Ki-67表达水平的价值。方法回顾性分析2015年1月~2024年6月在新疆医科大学第三临床医学院经病理学及免疫组化检查证实为TET的171例患者,按照7∶3的比例随机划分为训练组(n=120)和测试组(n=51)。对所有临床风险因素行Logistic回归分析,建立临床模型。利用ITK-SNAP勾画患者增强胸部CT薄层图像中的感兴趣区,利用Pyradiomics提取影像组学特征,采用Mann-Whitney U检验及LASSO回归进行特征筛选,采用二分类逻辑回归构建预测模型,并使用列线图可视化最佳模型。结果临床模型、影像组学模型及临床-影像组学联合模型在训练组的曲线下面积分别为0.670、0.867、0.890,测试组的曲线下面积分别为0.632、0.834、0.851。3组模型中临床-影像组学联合模型的预测效能最佳。结论基于增强CT影像组学特征及临床特征构建的模型在术前预测TET Ki-67表达水平中有一定的价值,且临床-影像组学联合模型优于临床模型及影像组学模型的单一模型。Objective To investigate the value of CT imaging histology combined with clinical features model for preoperative prediction of Ki-67 expression level in thymic epithelial tumors(TET).Methods A retrospective analysis of 171 patients with TET confirmed by pathology and immunohistochemistry at the Third Clinical Medical College of Xinjiang Medical University from January 2015 to June 2024 were randomly divided into a training group(n=120)and a testing group(n=51)according to a 7:3 ratio.Logistic regression analysis was performed for all clinical risk factors to create a clinical model.ITK-SNAP was used to outline the regions of interest in the enhanced chest CT thin-layer images of patients,Pyradiomics was used to extract the image histological features,Mann-Whitney U test and LASSO regression were used for feature screening,binary logistic regression was used to construct the prediction model,and the best model was visualized using a column-line diagram.Results The area under the curve of the clinical model,the imaging histology model,and the combined clinical-imaging histology model were 0.670,0.867,and 0.890 in the training group,and 0.632,0.834,and 0.851 in the test group,respectively;The combined clinical-imaging histology model had the best predictive efficacy among the three groups of models.Conclusion The model constructed based on enhanced CT imaging histologic features and clinical features has some value in predicting TET Ki-67 expression level before surgery,and the combined clinical-imaging histologic model is better than the clinical model and the single model of imaging histologic model.

关 键 词：胸腺上皮性肿瘤 影像组学 KI-67 

分 类 号：R736.3[医药卫生—肿瘤] R730.44[医药卫生—临床医学]