基于T_(2)WI图像构建的影像组学模型预测宫颈鳞状细胞癌分期的价值  

The value of T_(2)WI image-based radiomics model for predicting cervical squamous-cell carcinoma staging

作　　者：周永进 赵雪妙 钟屹 王海林[1] 孔春丽 纪建松[1] ZHOU Yongjin;ZHAO Xuemiao;ZHONG Yi;WANG Hailin;KONG Chunli;JI Jiansong(Department of Radiology,Lishui Central Hospital,Lishui,Zhej iang Province 323000,China)

机构地区：[1]丽水市中心医院放射科,浙江丽水323000

出　　处：《实用放射学杂志》2022年第3期428-431,共4页Journal of Practical Radiology

基　　金：浙江省医药卫生科技面上项目(2020KY1080).

摘　　要：目的探讨基于T_(2)WI联合临床指标建立的影像组学预测模型在鉴别宫颈鳞状细胞癌(CSCC)低分期(ES)和高分期(AS)的价值.方法回顾性分析181例CSCC患者的治疗前T_(2)WI图像和临床资料.采用AK软件对T_(2)WI图像提取纹理特征,使用最小绝对收缩与选择算子算法(LASSO)降维后建立鉴别CSCC患者ES和AS的影像组学标签.运用单因素Logistic回归筛选独立临床危险因素,采用多变量Logistic回归构建预测CSCC分期的列线图;采用受试者工作特征(ROC)曲线评估影像组学模型在训练组中的准确性,并通过验证组进行验证.采用校正曲线评估列线图预测和实际观察CSCC分期的一致性.结果影像组学标签、肿瘤最大径、年龄和鳞状细胞癌抗原(SCC-Ag)水平为独立危险因素.训练组和验证组中预测CSCC分期的曲线下面积(AUC)分别为0.873[95%置信区间(CI)0.812~0.935]和0.840(95%CI 0.725~0.955).校正曲线显示列线图在预测和实际观察之间具有良好的一致性.结论基于T_(2)WI图像联合临床指标建立的影像组学预测模型有助于鉴别CSCC患者ES与AS.Objective To investigate and develop a radiomics prediction model based on T_(2) WI texture parameters and clinical variables for differentiating the FIGO early stage(ES)and advanced stage(AS)cervical squamous-cell carcinoma(CSCC).Methods The pre-treatment T_(2) WI image data and clinical variables of 181 patients with CSCC confirmed by histopathology were analyzed retrospectively.AK software was used to extract the texture features of T_(2) WI images,and least absolute shrinkage and selection operator(LASSO)was used for feature reduction to establish a radiomics signature.UnivariateLogisticregression was used to screen independent clinical risk factors.MultivariateLogisticregression was used to establish a nomogram and radiomics prediction model with radiomics signature and clinical variables.Receiver operating characteristic(ROC)curve was used to evaluate the accurate of radiomics prediction model in training set and validation set.The calibration curve was used to evaluate the consistency between the nomogram prediction and the actual observation of CSCC staging.Results Radiomics signature,tumor diameter,age,and squamous cell carcinoma antigen(SCC-Ag)level were independent risk factor for the CSCC staging.The area under the curve(AUC)for predicting CSCC staging in training set and validation set was 0.873[95%confidence interval(CI)0.812-0.935]and 0.840(95%CI 0.725-0.955),respectively.The calibration curve showed that the nomogram had a good consistency between prediction and actual observation.Conclusion Radiomics prediction model based on T_(2) WI texture parameters and clinical variables are valuable in distinguishing FIGO ES and AS in CSCC patients.

关 键 词：宫颈鳞状细胞癌 T_(2)加权成像 影像组学 模型 

分 类 号：R737.33[医药卫生—肿瘤] R445.2[医药卫生—临床医学]