临床-影像组学诺模图在术前预测直肠癌淋巴结转移的应用价值

Application Value of Clinic-Radiomics Nomogram in Preoperative Prediction of Lymph Node Metastasis of Rectal Cancer

目的探讨联合多序列MRI影像组学和临床特征构建的诺模图在术前预测直肠癌淋巴结转移的价值。方法回顾性分析2018年12月至2020年10月蚌埠医学院第一附属医院术前行MRI检查且术后经病理证实的242例直肠癌患者的临床和影像资料。按照7∶3的比例随机将所有病例分为训练组(n=168)与验证组(n=74)。在训练组,对T2WI、增强T1WI(T1WI+C)和扩散加权成像(DWI)图像进行手动勾画病灶感兴趣区(ROI)并提取影像组学特征,经过数据降维,筛选出有效特征并构建影像组学预测模型。接着纳入相关临床危险因素,利用逻辑回归筛选出最有价值的临床危险因素并构建临床特征模型。采用受试者工作特征曲线(ROC)曲线下面积(AUC)来评价模型的诊断效能并利用最佳模型构建诺模图。决策曲线分析(DCA)和校正曲线评价诺模图的临床应用价值。结果联合临床特征和影像组学构建的临床-影像组学模型的诊断效能最高(AUC=0.86),均高于单一影像组学模型(AUC=0.84)和临床模型(AUC=0.62)。结论临床-影像组学诺模图在预测直肠癌术前淋巴结转移方面具有可行性。

Objective To explore the value of nomogram combined with multi-sequence MRI radiomics and clinical features in preoperative prediction of lymph node(LN)metastasis of rectal cancer.Methods From December 2018 to October 2020,the clinical and imaging data of 242 patients with rectal cancer confirmed by pathology and preoperative MRI examination in the First Affiliated Hospital of Bengbu Medical College were analyzed retrospectively.According to the ratio of 7:3,all cases were randomly divided into training group(n=168)and validation group(n=74).In the training set,T2WI,T1WI+C and diffusion weighted imaging(DWI)images were used to manually delineate the regions of interest and extract the radiomics features.After data dimensionality reduction,the effective features were selected and the radiomics prediction model was constructed.Then,logistic regression was used to screen out the most valuable clinical risk factors and construct the clinical characteristic model.The area under curve(AUC)of receiver operating characteristic(ROC)was used to evaluate the diagnostic efficiency of the model,and the best model was used to construct the nomogram.Decision curve analysis(DCA)and correction curve were used to evaluate the clinical value of nomogram.Results The diagnostic efficiency of clinic-radiomics model was the highest(AUC=0.86),which was higher than that of single radiomics model(AUC=0.84)and clinical model(AUC=0.62).Conclusion It is feasible to predict preoperative lymph node metastasis of rectal cancer by clinic-radiomics nomogram.