CT-Based Radiomics for Predicting the Therapeutic Effect of Hepatocellular Carcinoma After Transcatheter Arterial Chemoembolization
10.3969/j.issn.1005-5185.2025.03.006
- VernacularTitle: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
- Author:
Han YANG
1
;
Hao LI
1
;
Gengfei CAO
1
;
Yingjun BAO
1
;
Diwen ZHU
1
;
Junpeng GU
1
;
Weixin REN
1
Author Information
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
- Publication Type:Journal Article
- Keywords:
Carcinoma,hepatocellular;
Tomography,X-ray computed;
Transcatheter arterial chemoembolization;
Therapeutic effect;
Radiomics;
Forecasting
- From:
Chinese Journal of Medical Imaging
2025;33(3):252-259
- CountryChina
- Language:Chinese
-
Abstract:
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.
