临床-影像组学联合模型对直肠癌微卫星不稳定性的预测价值  

VALUE OF A CLINICAL-RADIOMICS COMBINED MODEL IN PREDICTING MICROSATELLITE INSTABILITY IN RECTAL CANCER

作　　者：刘旭[1] 冯卫华[1] 王涛[1] 高远翔[1] 李志明[1] LIU Xu;FENG Weihua;WANG Tao;GAO Yuanxiang;LI Zhiming(Department of Radiology, The Affiliated Hospital of Qingdao University, Qingdao 266003, China)

机构地区：[1]青岛大学附属医院放射科,山东青岛266003

出　　处：《精准医学杂志》2022年第4期326-331,共6页Journal of Precision Medicine

基　　金：青岛大学附属医院临床医学+X项目(QDFY+X202101017)。

摘　　要：目的探讨基于MRI影像组学特征和临床因素构建的联合模型在术前预测直肠癌微卫星不稳定性(MSI)的价值。方法回顾性分析2016年3月—2021年3月青岛大学附属医院术前行直肠MR检查且术后经病理检查证实的121例直肠癌患者的临床和影像学资料。所有病例按照7∶3的比例随机分为训练组(85例)与验证组(36例)。通过3D Slicer软件勾画三维感兴趣区并提取影像组学特征。利用最小绝对收缩和选择算子算法进行特征降维并构建影像组学模型。通过Logistic回归分析筛选临床因素并构建临床模型以及联合模型。绘制受试者工作特征(ROC)曲线评估临床模型、影像组学模型以及联合模型的预测效能。使用Delong检验比较3种模型的预测效能是否具有统计学差异。采用决策曲线评估3种模型的临床应用价值。结果共筛选8个影像组学特征用于构建影像组学模型,血液中高水平的血小板和低水平的高密度脂蛋白为直肠癌MSI的临床危险因素。联合模型对于直肠癌MSI显示出了较好的预测性能,训练组AUC为0.966,验证组AUC为0.931。Delong检验结果显示,在训练组和验证组中联合模型、影像组学模型的预测效能与临床模型比较,差异具有显著性(训练组Z=3.773、2.017,P<0.05;验证组Z=2.395、1.980,P<0.05)。决策曲线表明联合模型预测直肠癌MSI的净收益最大。结论基于MRI影像组学特征和临床因素构建的联合模型在术前可以有效预测直肠癌MSI。Objective To investigate the value of a combined model based on MRI radiomics features and clinical factors in predicting microsatellite instability(MSI)in rectal cancer before surgery.Methods A retrospective analysis was performed for the clinical and imaging data of 121 patients with rectal cancer who underwent preoperative rectal MR examination and were gi-ven a confirmed diagnosis by postoperative pathology in The Affiliated Hospital of Qingdao University from March 2016 to March 2021.All patients were randomly divided into training group(85 patients)and validation group(36 patients)at the ratio of 7∶3.3D Slicer software was used to delineate the three-dimensional region of interest and extract radiomics features.The least absolute shrinkage and selection operator algorithm was used to reduce feature dimension and construct a radiomics model.A logistic regression analysis was used to screen out clinical factors and construct the clinical model and the combined model.The receiver operating characteristic(ROC)curve was plotted to assess the predictive efficiency of the clinical model,the radiomics model,and the combined model.The Delong test was used to investigate whether there was a statistical difference in predictive performance between the three models.Decision curve analysis was used to assess the clinical application value of the three models.Results A total of 8 radiomics features were screened out to construct the radiomics model,and the high level of platelet and the low level of highdensity lipoprotein in blood were the clinical risk factors for MSI in rectal cancer.The combined model showed a good predictive performance for MSI,with an area under the ROC curve of 0.966 in the training group and 0.931 in the validation group.The Delong test showed that in both the training group and the validation group,there was a significant difference in predictive performance between the combined model and the clinical model,as well as between the radiomics model and the clinical model(training group:Z=3.773,2.

关 键 词：直肠肿瘤 磁共振成像 成像 三维 微卫星不稳定性 影像组学 诊断效能 ROC曲线 

分 类 号：R735.37[医药卫生—肿瘤] R445[医药卫生—临床医学]