摘要
目的利用增强CT影像组学特征构建影像组学模型并绘制诺莫图鉴别小肝癌与肝不典型增生结节。方法收集经病理证实为小肝癌(n=60)与肝不典型增生结节(n=57)患者的临床资料及CT影像数据。于动脉期、门脉期、延迟期CT图像手动勾画肝脏病灶ROI,提取组学特征后行特征降维和筛选,于筛选出的特征中分别选取1-10个特征建立逻辑回归(LR)、支持向量机(SVM)、决策树(DT)模型,利用5倍交叉进行内部验证,以交叉验证集AUC值最高的模型作为最优模型。构建影像组学评分公式并绘制诺莫图。行拟合优度检验评估模型的拟合度,绘制决策曲线评价其净获益。结果最优模型为LR模型,其内部交叉验证AUC值为0.795(95%CI 0.644-0.880),训练组AU C值为0.860(95%CI 0.772-0.936),测试组AU C值为0.807(95%CI 0.650-0.952),模型的校正曲线具有良好的一致性(p=0.970,P>0.05),决策曲线也具有较高的净获益。结论利用肝脏增强CT组学特征构建的影像组学模型能够有效鉴别小肝癌和肝不典型增生结节。
Objective Constructing a radiomics model and nomogram using contrast-enhanced CT radiomics features to differentiate small hepatocellular carcinoma from hepatic dysplastic nodules.Methods The clinical data and CT imaging data of patients with pathologically confirmed s-HCC(n=60)and HDN(n=57)were collected.The ROI of liver lesions was manually sketched on the CT images of the arterial phase,portal phase,and delayed phase.Radiomics features were extracted,and then feature dimension reduction and screening were performed.Among the selected features,1-10 features were selected respectively to establish logistic regression(LR),support vector machine(SVM)and decision tree(DT)models,and fivefold cross-validation was used for internal verification.The model with the highest AUC value in the crossvalidation set was taken as the optimal model.Build a radiomics scoring formula and draw a nomogram.A goodness-of-fit test was performed to evaluate the fit degree of the model,and a decision curve was drawn to evaluate its net benefit.Results The optimal model is the LR model with the internal crossvalidation AUC value of 0.795(95%CI 0.644-0.880),the training group AUC value of 0.860(95%CI 0.772-0.936),and the test group AUC value of 0.807(95%CI 0.650-0.952),the calibration curve of the model has good agreement(P=0.970,P>0.05),and the decision curve also has a high net benefit.Conclusion A radiomics model constructed using the radiomics features of liver-enhanced CT can be used as an effective tool for the identification of small hepatocellular carcinoma and hepatic dysplastic nodules.
作者
邢艳虎
黄忠江
侯跃宏
XING Yan-hu;HUANG Zhong-jiang;HOU Yue-hong(Department of Imaging,Shanxi Cardiovascular Hospital,Taiyuan 030000,Shanxi Province,China;Department of Imaging,Shanxi Provincial Hospital of Traditional Chinese Medicine,Taiyuan 030000,Shanxi Province,China)
出处
《中国CT和MRI杂志》
2023年第7期107-109,共3页
Chinese Journal of CT and MRI
关键词
小肝癌
肝不典型增生结节
计算机断层扫描
影像组学
Small hepatocellular Carcinoma
Hepatic Dysplastic Nodule
Computed Tomography
Radiomics
