放射学实践2024，Vol.39Issue(5) ：585-590.DOI:10.13609/j.cnki.1000-0313.2024.05.004

MRI影像组学联合炎症因子对肝细胞肝癌微血管侵犯的预测价值

The predictive value of MRI radiomics integrated with inflammatory factors for microvascular invasion in hepatocellular carcinoma

杨砾寒 陈梦林 陈诗 赵新湘
放射学实践2024，Vol.39Issue(5) ：585-590.DOI:10.13609/j.cnki.1000-0313.2024.05.004
✕
  • 维普
  • 万方数据

MRI影像组学联合炎症因子对肝细胞肝癌微血管侵犯的预测价值

The predictive value of MRI radiomics integrated with inflammatory factors for microvascular invasion in hepatocellular carcinoma

杨砾寒 1陈梦林 1陈诗 1赵新湘1
扫码查看

作者信息

  • 1. 650032 昆明,昆明医科大学第二附属医院放射科
  • 折叠

摘要

目的:探讨基于MRI影像组学联合炎症因子术前预测肝细胞肝癌(HCC)微血管侵犯(MVI)的价值.方法:纳入经病理证实的HCC患者221例,其中MVI阳性117例,MVI阴性104例.比较MVI阴性与阳性患者的炎症因子、影像特征差异,运用多因素Logistic分析确定MVI的独立危险因素,建立影像特征及炎症因子预测模型.勾画Gd-DTPA增强门静脉期瘤周20 mm及瘤内所有层面,使用最小绝对收缩和选择算子(LASSO)算法筛选影像组学特征,建立瘤周、瘤内、瘤周及瘤内共三种影像组学模型.选择瘤周、瘤内影像组学及炎症因子建立联合预测模型,使用ROC曲线在验证组中评估模型的预测效能.结果:Logistic多因素分析结果显示肿瘤最大直径、包膜、动脉期瘤周强化、[碱性磷酸酶(ALP)+γ-谷氨酰转肽酶(GGT)]/淋巴细胞计数(AGLR)是MVI的独立危险因素,基于上述独立危险因素建立的影像特征及炎症因子预测模型预测HCC MVI的ROC曲线下面积(AUC)训练组为0.80,验证组为0.75.基于瘤周及瘤内影像组学建立的影像组学模型较仅包含瘤内影像组学的模型预测HCC MVI的AUC高(瘤周及瘤内模型在训练组和验证组的AUC分别为0.83、0.79,瘤内模型在训练组和验证组的AUC分别为0.75、0.73).瘤周、瘤内影像组学及炎症因子构建的联合预测模型预测HCC MVI的AUC训练组为0.87,验证组为0.82.结论:基于Gd-DTPA门静脉期建立的瘤周及瘤内影像组学模型可对HCC MVI进行术前预测,联合炎症因子可进一步提高其预测效能.

Abstract

Objectives:To investigate the value of MRI radiomics integrated with inflammatory factors in predicting microvascular invasion(MVI)of hepatocellular carcinoma(HCC)preoperatively.Methods:A total of 221 patients with pathologically confirmed HCC were included,including 117 MVI-positive and 104 MVI-negative cases.After univariate analysis of inflammatory factors and imaging features,valid predictors of MVI were identified by using multivariate logistic regression analysis.These predictors were then applied to establish a predictive model.All layers of the tumors(intratu-moral and peritumoral 20mm)were outlined.The least absolute shrinkage and selection operator(LASSO)was used to select the radiomics signatures,and three radiomic models were established:peritumoral,intratumoral,peri-and intratumoral models.Inflammatory factors were combined with peritumoral and intratumoral radiomics signatures to produce a combined model.The predictive effi-ciency of the models were validated in the testing groups using ROC curves.Results:Multivariate logis-tic regression analysis showed maximum tumor diameter,capsule,peritumoral hyperintensity in the arterial phase,and[alkaline phosphatase(ALP)+gamma-glutamyl transpeptidase(GGT)]/lympho-cyte count ratio(AGLR)were independent risk factors for MVI.Based on the above parameters,the imaging characteristics and inflammatory factors were modelled.The area under the ROC curve(AUC)of this model was 0.80 and 0.75 for the training and testing groups,respectively.The AUC of the peritumoral and intratumoral radiomics model was higher than intratumoral radiomics model(training group:0.83 vs 0.75;testing group:0.79 vs 0.73).The AUC of combined predictive model,built by intratumoral and peritumoral radiomics signatures and inflammatory factors,were 0.87 for the training group and 0.82 for the testing group.Conclusion:The peritumoral and intratumoral radiomics model based on the GD-DTPA portal phase can be used for preoperatively predicting MVI in HCC.Ad-ditionally,the inflammatory factor might further enhance the predictive efficacy of this model.

关键词

肝细胞肝癌/影像组学/磁共振成像/微血管侵犯/炎症指标/影像特征

Key words

Hepatocellular carcinoma/Radiomics/Magnetic resonance imaging/Microvascular invasion/Inflammatory indicators/Imaing features

引用本文复制引用

出版年

2024
放射学实践
华中科技大学同济医学院

放射学实践

CSTPCDCSCD北大核心
影响因子：1.08
ISSN：1000-0313
参考文献量23
段落导航相关论文