基于术后CT影像组学预测肝细胞癌经动脉化疗栓塞后的肿瘤反应  

CT-Based Radiomics for Predicting the Therapeutic Effect of Hepatocellular Carcinoma After Transcatheter Arterial Chemoembolization

作　　者：杨瀚 李毫 曹耿飞 鲍应军 朱帝文[1] 顾俊鹏[1] 任伟新[1] YANG Han;LI Hao;CAO Gengfei;BAO Yingjun;ZHU Diwen;GU Junpeng;REN Weixin(Department of Interventional Radiology,the First Affiliated Hospital of Xinjiang Medical University,Urumqi 830054,China)

机构地区：[1]新疆医科大学第一附属医院介入放射科,新疆乌鲁木齐830054

出　　处：《中国医学影像学杂志》2025年第3期252-259,共8页Chinese Journal of Medical Imaging

基　　金：新疆维吾尔自治区自然科学基金重点项目(2021D01D22)。

摘　　要：目的基于术后CT影像组学识别重复经导管动脉化疗栓塞(TACE)有反应的肝细胞癌的价值。资料与方法回顾性收集2018年2月—2024年5月于新疆医科大学第一附属医院确诊的肝细胞癌96例,患者均接受3次或以上TACE治疗,以7∶3随机分为训练组67例和验证组29例。运用3D-Slicer软件阈值分割第一次TACE术后24h内复查CT图像靶病灶中碘油沉积区,包括平扫和动脉期,最终筛选出5个影像组学特征,分别构建影像组学预测模型以识别重复TACE后有反应的肝细胞癌。采用受试者工作特征曲线及决策曲线评价模型的预测效能。结果Child-Pugh分级(OR=2.737,P<0.05)、巴塞罗那临床肝癌分期(OR=2.907,P<0.05)、肿瘤数量多发(OR=4.505,P<0.01)和甲胎蛋白(OR=1.002,P<0.01)是预测TACE术后肿瘤反应的独立危险因素。训练组中,平扫和动脉期模型的曲线下面积分别为0.813、0.831,验证组中分别为0.748、0.788。两种模型均能有效评估肝细胞癌TACE后的肿瘤反应。结论基于TACE术后首次CT影像组学特征建立的预测模型可以有效识别经重复TACE后有反应的肝细胞癌,靶病灶内碘油沉积模式可作为影像生物标志物评估TACE术后肿瘤反应的价值。Purpose To identify hepatocellular carcinoma that responds to repetitive transcatheter arterial chemoembolization(TACE)based on CT radiomics.Materials and Methods A total of 96 patients diagnosed as hepatocellular carcinoma in the First Affiliated Hospital of Xinjiang Medical University from February 2018 to May 2024 were randomly divided into a training group(n=67)and internal validation group(n=29)at a ratio of 7∶3.All patients received three or more TACE treatments.Radiomics features were extracted from lipiodol of the target-lesions by semi-automatic segmentation on the axial CT image after TACE within 24 hours.The radiomics model were constructed by five features for differentiating non-response group from response group.Receiver operating characteristic curve analysis and decision curve were performed to evaluate the performance of the model.Results Child-Pugh classification(OR=2.737,P<0.05),BCLC stage(OR=2.907,P<0.05),multifocal tumors(OR=4.505,P<0.01)and alpha-fetoprotein(OR=1.002,P<0.01)were independent risk factors for predicting tumor response after TACE.The area under the curve of the non-contrast CT based model and the arterial-enhanced CT based model were 0.813 and 0.831 in the experimental group;0.748 and 0.788 in the validation group,respectively.Both of the two models showed good prediction performance.Conclusion The radiomics model based on CT imaging features after first TACE is effectively for differentiating non-response group from response group,lipiodol retention patterns from the target lesion can be the imaging biomarkers for TACE response prediction.

关 键 词：癌 肝细胞 体层摄影术 X线计算机 经导管动脉化疗栓塞 肿瘤反应 影像组学 预测 

分 类 号：R445.3[医药卫生—影像医学与核医学] R735.7[医药卫生—诊断学]