摘要
目的探讨基于轴位脂肪抑制T2加权成像(FS-T2WI)及T1加权增强(T1CE)序列联合临床特征创建的临床-影像组学模型术前预测直肠癌淋巴结转移的应用价值。资料与方法回顾性分析2016年12月—2020年12月河南省人民医院收治的行MRI且术后病理证实为直肠癌的350例患者的临床资料,比较淋巴结转移组(n=131)及无转移组(n=219)的一般特征(年龄、性别、癌胚抗原、糖类抗原199、最大直径、有无转移)并筛选出差异有统计学意义的临床危险因素;手动勾画肿瘤全层感兴趣区,利用开源软件PyRadiomics提取影像组学特征,并经过特征降维构建影像组学分数(Rad-score),联合临床危险因素,构建临床-影像组学预测模型(共6个,分别为FS-T2WI、T1CE、FS-T2WI+T1CE、FS-T2WI联合临床因素、T1CE联合临床因素及FS-T2WI+T1CE联合临床因素)。以7∶3将患者随机分为训练集(245例)和测试集(105例),根据受试者工作特征曲线评估不同模型的预测效能。结果癌胚抗原值及有无远处转移是发生淋巴结转移的危险因素(t=2.570、χ2=11.234,均P<0.001)。FS-T2WI、T1CE及FS-T2WI+T1CE联合临床特征模型的曲线下面积(AUC)分别显著大于单纯FS-T2WI、T1CE及两者联合序列模型的AUC(0.854比0.718,Z=2.951;0.925比0.888,Z=5.556;0.942比0.907,Z=2.779;P均<0.005),FS-T2WI+T1CE与临床联合模型的AUC大于FS-T2WI与临床联合模型的AUC(0.942比0.854,Z=4.061,P<0.001),T1CE与临床联合模型的AUC大于FS-T2WI与临床联合模型的AUC(0.925比0.854,Z=3.733,P=0.0002)。结论本研究构建的临床-影像组学模型对术前个体化预测直肠癌患者淋巴结转移具有较高的价值。
Purpose To explore the application value of creating an clinical-radiomics model to predict preoperative lymph node metastasis in patients with rectal cancer based on fat suppression-T2 weighted imaging(FS-T2WI)and T1-weighted contrast-enhanced(T1CE)sequences combined with clinical features.Materials and Methods The clinical data of 350 patients with rectal cancer who underwent MRI and were confirmed by postoperative pathology in Henan Provincial People's Hospital from December 2016 to December 2020 were retrospectively analyzed.The general characteristics[age,gender,carcinoembryonic antigen(CEA),carbohydrate antigen 199,maximum diameter,and metastasis]of the lymph node metastasis group(n=131)and the non-metastasis group(n=219)were compared,and the clinical risk factors with statistically significant differences were screened out.The full-thickness tumor region of interest was manually delineated,and the radiomics features were extracted using the open source software PyRadiomics,and the radiomics score(Rad-score)was constructed after feature dimension reduction.Combined with clinical risk factors,clinical-radiomics prediction models were constructed(a total of 6,FS-T2WI,T1CE,FS-T2WI+T1CE,FS-T2WI combined with clinical factors,T1CE combined with clinical factors and FS-T2WI+T1CE combined with clinical factors).The patients were randomly divided into a training set(245 cases)and a test set(105 cases)at a ratio of 7∶3.The predictive efficacy of different models was evaluated according to the receiver operating characteristic curve.Results The CEA value and distant metastasis were risk factors for lymph node metastasis(t=2.570,χ2=11.234,all P<0.001).The area under curve(AUC)of FS-T2WI,T1CE and FS-T2WI+T1CE combined with clinical features model was significantly higher than that of FS-T2WI,T1CE and their combination sequence model(0.854 vs.0.718,Z=2.951;0.925 vs.0.888,Z=5.556;0.942 vs.0.907,Z=2.779;all P<0.005).The AUC of FS-T2WI+T1CE combined with clinical model was better than that of FS-T2WI combined with clini
作者
杨燕
付芳芳
吴亚平
吉祥
王梅云
YANG Yan;FU Fangfang;WU Yaping;JI Xiang;WANG Meiyun(Department of Medical Imaging,the People's Hospital of Zhengzhou University,Zhengzhou 450003,China;Henan Provincial People's Hospital,Henan Key Laboratory of Neurological Imaging,Zhengzhou 450003,China;不详)
出处
《中国医学影像学杂志》
CSCD
北大核心
2022年第9期889-895,共7页
Chinese Journal of Medical Imaging
基金
河南省科技公关项目(212102310689)
河南省医学科技攻关计划项目(LHGJ20210001,LHGJ20210005)。
关键词
直肠肿瘤
磁共振成像
淋巴结转移
影像组学
病理学
外科
预测
Rectal tumors
Magnetic resonance imaging
Lymphatic metastasis
Radiomics
Pathology,surgical
Prognosis
