Multiphase Enhanced CT-Based Radiomics for Predicting Recurrence in Patient with Hepatocellular Carcinoma After Tumor Resection
10.3969/j.issn.1005-5185.2025.03.005
- VernacularTitle:基于多期增强CT影像组学模型预测肝细胞癌切除术后复发的价值
- Author:
Chunyan YANG
1
;
Xiaoqin WEI
;
Qiong YANG
;
Yang LI
;
Yong DU
Author Information
1. 川北医学院附属医院放射科,四川 南充 637000;武胜县人民医院医学影像科,四川 广安 638400
- Publication Type:Journal Article
- Keywords:
Carcinoma,hepatocellular;
Recurrence;
Tomography,X-ray computed;
Tumor resection;
Radiomics;
Risk factors
- From:
Chinese Journal of Medical Imaging
2025;33(3):245-251
- CountryChina
- Language:Chinese
-
Abstract:
Purpose To develop and validate a radiomics-clinical model that could accurately predict the recurrence of hepatocellular carcinoma(HCC)after undergoing tumor resection.Materials and Methods A total of 311 HCC patients underwent tumor resection in the Affiliated Hospital of North Sichuan Medical College from January 2015 to June 2022 were retrospectively collected,and they were randomly divided into a training cohort(n=217)and a validation cohort(n=94)in the ratio of 7∶3.Tumor and peritumoral 5 mm regions of interest were outlined on arterial and portal venous phase images and radiomics features were extracted to establish an arterio-portal radiomics model.Independent clinical risk factors associated with postoperative recurrence were explored by univariate and multivariate Cox analysis,then clinical model was established.A combined radiomics-clinical model was established by combining clinical independent risk factors and radiomics features.The area under the curve,specificity,sensitivity,net reclassification improvement and integrated discrimination improvement were used to assess the discrimination of all models.The calibration curve assessed the calibration of the model and decision curve analysis assessed the clinical utility of the model.Validation was performed by validation cohort data.Results The Rad-A5V5-clinical model had the best predictive performance for postoperative recurrence of HCC,the area under the curve,specificity and sensitivity of the validation cohort were 0.743(95%CI 0.640-0.846),0.647 and 0.717,respectively.The results of net reclassification improvement and integrated discrimination improvement showed that compared with clinical model and radiomics model,the prediction ability of Rad-A5V5-clinical model was improved the best one.The calibration curves showed that the predicted values of the Rad-A5V5-clinical model conformed the most favorably to the true values.The results of the decision curve analysis curve analyses showed that,among all models,the Rad-A5V5-clinical model would obtain the largest net benefit within a certain threshold range.Conclusion The predictive efficacy for post-tumor resection recurrence in HCC is significantly improved by incorporating tumor-peritumor radiomics features along with clinically independent risk factors.This finding offers a crucial reference point for identifying at-risk patients and tailoring individualized treatment plans.
