基于增强CT图像纹理特征术前预测肝细胞癌微血管侵犯状态及危险度等级  

Contrast-enhanced CT radiomics texture analysis for preoperative prediction of microvascular invasion state and risk in hepatocellular carcinoma

作　　者：潘克华[1] 张昭[1] 贾秀芬[1] 刘瑾瑾[1] 陈永华[1] PAN Ke-hua;ZHANG Zhao;JIA Xiu-fen(Department of Radiology,the First Affiliated Hospital of Wenzhou Medical University,Zhejiang 325000,China)

机构地区：[1]温州医科大学附属第一医院放射科,浙江325000

出　　处：《放射学实践》2023年第9期1096-1100,共5页Radiologic Practice

基　　金：温州市科技局科技计划项目(Y2020170)。

摘　　要：目的:探讨基于增强CT图像纹理特征模型术前预测肝细胞癌(HCC)微血管侵犯(MVI)的价值。方法:回顾性搜集本院2018年1月至2022年12月经手术病理证实的HCC患者496例,按2:1的比例随机分为训练组(331例)和测试组(165例)。采用ITK SNAP图像纹理分析软件对HCC瘤灶及瘤周邻近区域勾画兴趣区(ROI)并进行图像纹理特征提取、筛选,采用最小绝对收缩与选择算子(LASSO)回归算法对576个纹理特征进行降维,使用多变量Logistic回归提取有意义的纹理特征建立模型以预测MVI状态及危险度等级。联合纹理特征和肿瘤临床分期建立列线图以预测MVI危险度等级。采用ROC曲线下面积(AUC)评价模型的诊断效能。结果:训练组与测试组患者的年龄、性别、肿瘤位置差异均无统计学意义。基于增强CT图像纹理特征模型可以较好地预测MVI状态及危险度等级,在训练组和验证组中预测有无MVI的AUC分别为0.783、0.773,敏感度分别为0.705、0.883,特异度分别为0.750、0.722;在训练组和验证组中预测MVI危险度等级的AUC分别为0.743、0.718。联合纹理特征和肿瘤临床分期建立的列线图对MVI危险度等级的预测效能(AUC=0.856)优于单纯纹理特征模型。结论:基于增强CT图像纹理特征模型可用于术前预测肝细胞癌的MVI状态和危险度等级,是一种可靠的临床评估工具,对临床医师选择合适的治疗方案、准确评估预后具有重要参考价值。Objective:To investigate the prognostic value of texture analysis of contrast-enhanced computed tomography(CT)imaging in preoperatively predicting microvascular invasion(MVI)status(positive vs negative)and risk(low vs high)in patients with hepatocellular carcinoma(HCC).Methods:A total of 496 pathology-proven HCC patients were retrospectively included from January 2018 to December 2022.The patients were randomly divided into a training cohort of 331 patients and a test cohort of 165 patients with the ratio of 2:1.The tumor lesion and peri-tumoral area was segmented manually on the largest cross-sectional slice with ITK SNAP according to agreement between two radiologists.The LASSO algorithm was used for the selection of 576 radiomics features.Two classifiers for predicting MVI status and MVI risk were developed using multivariable logistic regression.The receiver operating characteristic curve(ROC)and the area under curve(AUC)were used to evaluate the diagnostic efficiency of the models.Results:There were no significant statistical differences in age,sex,tumor location between training and test cohorts.The developed radiomics signature predicted MVI status with AUC of 0.783 in the training cohort and 0.773 in the test cohort.For MVI risk stratification,the AUCs of the radiomics signature were 0.743,0.718 in the training and test cohorts,respectively,and the AUC of the final MVI risk classifier-integrated clinical stage were 0.856.Conclusions:CT radiomics-based models can be used to predict MVI status and MVI risk of HCC,and may serve as a reliable preoperative evaluation tool.

关 键 词：肝细胞癌 体层摄影术 X线计算机 影像组学 微血管侵犯 

分 类 号：R735.7[医药卫生—肿瘤] R814.42[医药卫生—临床医学]