摘要
目的以X线计算机体层摄影术(CT)为基础,构建全肝影像组学模型,并对乙型肝炎肝纤维化分期进行评估。方法收集100例乙型肝炎肝脏纤维化患者的临床及影像资料,按7:3比例分为训练组与验证组,训练组71例,验证组29例。应用AK分析软件提取患者肝脏平扫图像的纹理属性,然后对训练组提取纹理特征展开特征降维操作,再构建出影像组学纤维化指数(RFI)模型。基于RFI模型使用ROC曲线分别用于评估不同纤维化分期的诊断效能,并评价其在验证组中的效能。结果降维后共剩余8个特征用于构建RFI模型,ROC曲线分析显示RFI模型在训练组和测试组中均表现出良好的预测效能,AUC分别为0.81和0.80,特异度分别为0.74和0.71,敏感度分别为0.81和0.75。结论应用肝脏CT构建出来的影像组学模型,能够对肝纤维化分期进行定量分析,有望为临床提供一种无创性评价工具。
Objective Based on X-ray computed tomography(CT),a whole liver imaging model was constructed,and the value of staging liver fibrosis in hepatitis B was evaluated.Methods The clinical and imaging data of 100 patients with hepatitis B liver fibrosis were selected retrospectively.The training group and verification group were randomly divided according to the ratio of 7:3,with the former 71 cases and the latter 29 cases.AK analysis software was used to extract the texture attributes of each patient's liver plain scan image,then the texture features of the training group were extracted,the feature dimensionality reduction operation was carried out,and then the image omics fibrosis index(RFI)model was constructed.Based on RFI model,ROC curve was used to evaluate the diagnostic efficacy of different fibrosis stages,and to evaluate its efficacy in the validation group.Results After dimensionality reduction,a total of 8 features were used to construct the RFI model.According to the results of ROC curve,RFI model showed good prediction efficiency in the training group and the test group.The AUC was 0.81 and 0.8,the specificity was 0.74 and 0.71,and the sensitivity was 0.81 and 0.75,respectively.Conclusion The imaging omics model constructed by liver CT could quantitatively analyze the stages of liver fibrosis.This non innovative evaluation tool is conducive to clinical application.
出处
《浙江临床医学》
2023年第7期975-977,共3页
Zhejiang Clinical Medical Journal
关键词
肝纤维化
影像组学
预测模型
肝脏穿刺
Hepatic fibrosis
Imaging omics
Prediction model
Liver puncture
