Prediction of anticoagulant treatment of portal vein thrombosis based on clinical and CT radiomics
10.3969/j.issn.1002-1671.2025.07.016
- VernacularTitle:基于临床及CT影像组学预测门静脉血栓抗凝治疗
- Author:
Peng LIU
1
;
Jingxuan ZHANG
1
;
Hui XU
1
;
Dawei YANG
1
;
Zhenghan YANG
1
Author Information
1. 首都医科大学附属北京友谊医院放射科,北京 100050
- Publication Type:Journal Article
- Keywords:
computed tomography;
radiomics;
portal vein thrombosis;
anticoagulation;
efficacy
- From:
Journal of Practical Radiology
2025;41(7):1153-1157
- CountryChina
- Language:Chinese
-
Abstract:
Objective To establish and validate a machine learning model integrating abdominal contrast-enhanced CT radiomics features and clinical characteristics,and to construct a predictive model for the efficacy of anticoagulant treatment in portal vein thrombosis(PVT).Methods A retrospective selection was conducted on 94 PVT patients who received anticoagulant treatment.Patients were divided into effective and ineffective treatment groups based on the follow-up results.Clinical information was collected,and imaging features were evaluated.Univariate and multivariate logistic regression were performed to select clinical information and imaging fea-tures for constructing a clinical-imaging model.On CT venous phase images,the PVT mask was delineated and radiomics features were extracted,and the radiomics model was screened and established.A combined model was further developed using features from both the clinical-imaging and radiomics models.Receiver operating characteristic(ROC)curves were used to evaluate the predictive efficacy of different models.Results The area under the curve(AUC)for the clinical-imaging model,radiomics model,and com-bined model were 0.594,0.794,and 0.776,respectively.The radiomics and combined models demonstrated superior predictive efficacy for anticoagulant treatment in PVT compared to the clinical-imaging model.No significant difference in performance was observed between radiomics and combined models.Conclusion The radiomics model and combined model based on abdominal contrast-enhanced CT can effectively predict the efficacy of anticoagulant treatment for PVT.
