摘要： 目的 探讨基于CT影像组学特征建立的模型对肿瘤突变负荷(TMB)状态的预测价值。方法 从癌症影像数据库(TCIA)下载37例肺磷状细胞癌(简称鳞癌)患者资料,在首次CT图像中勾画感兴趣区,再用医学图像处理软件(3D Slicer)提取影像组学特征并筛选出与TMB高度相关的特征,构建影像组学特征预测模型,进一步构建nomogram图,并评价其预测效能。结果 通过LASSO回归共筛选出与TMB相关的9个影像学特征,采用logistic回归,进一步筛选出TMB预测模型的3个影像学特征。该模型曲线下面积(AUC)为0.882,提示模型的预测效能良好。结论 基于CT影像组学特征建立的模型对肺鳞癌TMB状态有较好的预测价值。

构建CT影像组学模型预测肺鳞状细胞癌肿瘤突变负荷

管超^1,2, 李梦玲³, 蒋礼青¹, 李思敏³, 张宇冲¹, 张博³, 李智¹, 赵明芳¹

1. 中国医科大学附属第一医院肿瘤内科, 沈阳 110001;
2. 中国医科大学附属盛京医院肿瘤科, 沈阳 110004;
3. 中国医科大学附属第一医院临床流行病学与循证医学教研室, 沈阳 110001

收稿日期:2021-09-28出版日期:2022-06-30发布日期:2022-06-09
通讯作者:赵明芳E-mail:zhaomf618@126.com
作者简介:管超(1988-),女,主治医师,博士研究生.
基金资助:国家重点研发计划项目(2016YFC1303800)


关键词: 肺鳞状细胞癌, 影像组学, 肿瘤突变负荷, 预测
Abstract: Objective To investigate the predictive value of a model based on computed tomography (CT) histological features for tumor mutational burden(TMB). Methods The data of 37 patients with squamous cell lung cancer were obtained from The Cancer Imaging Archive,and the regions of interest were outlined in the first CT images. The imaging histological features were extracted using a medical image processing software(3D Slicer),and those highly correlated with TMB were screened to construct an imaging histological feature prediction model. Moreover,the nomogram was further constructed to evaluate its predictive efficiency. Results Nine imaging histological features associated with TMB were screened by LASSO regression,and three imaging features were further screened by logistic regression for the TMB prediction model. The area under the curve value of the model was 0.882,suggesting good predictive efficacy. Conclusion The model based on CT imaging histological features has good predictive value for TMB in squamous cell lung cancer.
Key words: squamous cell lung cancer, imaging histology, tumor mutational burden, prediction
PDF全文下载地址:

https://journal.cmu.edu.cn/CN/article/downloadArticleFile.do?attachType=PDF&id=3002