A machine learning model based on CT radiomics for preoperatively differentiating intrahepatic mass-type cholangiocarcinoma and inflammatory pseudotumours

Abstract Objective Intrahepatic cholangiocarcinoma (ICC) and hepatic inflammatory pseudotumours (IPTs) share similar imaging features, leading to unnecessary biopsies and surgeries. Accurate preoperative differentiation is essential. Current studies using traditional imaging analysis have limited ac...

Full description

Saved in:
Bibliographic Details
Main Authors: Xiao-chun Wang, Jing-hong Liang, Xiao-yao Huang, Wen-jian Tang, Yan-mei He, Jun-yuan Zhong, Ling Zhang, Lun Lu
Format: Article
Language:English
Published: BMC 2025-07-01
Series:BMC Cancer
Subjects:
Online Access:https://doi.org/10.1186/s12885-025-14488-z
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849399620736647168
author Xiao-chun Wang
Jing-hong Liang
Xiao-yao Huang
Wen-jian Tang
Yan-mei He
Jun-yuan Zhong
Ling Zhang
Lun Lu
author_facet Xiao-chun Wang
Jing-hong Liang
Xiao-yao Huang
Wen-jian Tang
Yan-mei He
Jun-yuan Zhong
Ling Zhang
Lun Lu
author_sort Xiao-chun Wang
collection DOAJ
description Abstract Objective Intrahepatic cholangiocarcinoma (ICC) and hepatic inflammatory pseudotumours (IPTs) share similar imaging features, leading to unnecessary biopsies and surgeries. Accurate preoperative differentiation is essential. Current studies using traditional imaging analysis have limited accuracy. We developed a machine learning model based on clinical and CT radiomic features to improve diagnostic accuracy. Methods From May 2008 to January 2024, the data of 112 patients with ICC and 34 patients with hepatic IPTs who underwent preoperative plain and enhanced CT scans and whose diseases were confirmed by surgery and pathology were retrospectively analysed. A radiomic feature set, a clinical feature set, and a radiomic + clinical feature set were developed, and each was used to construct 14 machine learning models. The optimal hyperparameters were identified using fivefold cross-validation and a grid search. Finally, the area under the curve (AUC), accuracy, recall, precision, F1, Kappa value and other indicators were used to evaluate the performance of the models in the test sets to determine the optimal model for each feature subset. Results The machine learning model constructed with the radiomic features of all the CT sequences and the fused model constructed with both clinical features + all the CT sequence radiomic features performed well (AUC = 0.91 and 0.97, respectively), whereas the performance of the machine learning model constructed with the clinical features alone was relatively poor (AUC = 0.73). In terms of model performance in identifying the two diseases, the accuracy of the fused model was better in identifying ICCs than in identifying IPTs. Conclusion A diagnostic model constructed from clinical and CT radiomic features quickly differentiated between IPT from ICC. The model may be helpful for the preoperative identification of IPTs and ICC.
format Article
id doaj-art-791359196dcd4c9387b8b2cbb24abd79
institution Kabale University
issn 1471-2407
language English
publishDate 2025-07-01
publisher BMC
record_format Article
series BMC Cancer
spelling doaj-art-791359196dcd4c9387b8b2cbb24abd792025-08-20T03:38:18ZengBMCBMC Cancer1471-24072025-07-012511910.1186/s12885-025-14488-zA machine learning model based on CT radiomics for preoperatively differentiating intrahepatic mass-type cholangiocarcinoma and inflammatory pseudotumoursXiao-chun Wang0Jing-hong Liang1Xiao-yao Huang2Wen-jian Tang3Yan-mei He4Jun-yuan Zhong5Ling Zhang6Lun Lu7Department of Medical Imaging, Ganzhou Institute of Medical Imaging, Ganzhou Key Laboratory of Medical Imaging and Artificial Intelligence, Medical Imaging Center, Ganzhou People’s Hospital, The Affiliated Ganzhou Hospital of Nanchang University, Ganzhou Hospital-Nanfang Hospital, Southern Medical UniversityDepartment of Medical Imaging, Ganzhou Institute of Medical Imaging, Ganzhou Key Laboratory of Medical Imaging and Artificial Intelligence, Medical Imaging Center, Ganzhou People’s Hospital, The Affiliated Ganzhou Hospital of Nanchang University, Ganzhou Hospital-Nanfang Hospital, Southern Medical UniversityDepartment of Medical Imaging, Ganzhou Institute of Medical Imaging, Ganzhou Key Laboratory of Medical Imaging and Artificial Intelligence, Medical Imaging Center, Ganzhou People’s Hospital, The Affiliated Ganzhou Hospital of Nanchang University, Ganzhou Hospital-Nanfang Hospital, Southern Medical UniversityDepartment of Medical Imaging, Ganzhou Institute of Medical Imaging, Ganzhou Key Laboratory of Medical Imaging and Artificial Intelligence, Medical Imaging Center, Ganzhou People’s Hospital, The Affiliated Ganzhou Hospital of Nanchang University, Ganzhou Hospital-Nanfang Hospital, Southern Medical UniversityDepartment of Medical Imaging, Ganzhou Institute of Medical Imaging, Ganzhou Key Laboratory of Medical Imaging and Artificial Intelligence, Medical Imaging Center, Ganzhou People’s Hospital, The Affiliated Ganzhou Hospital of Nanchang University, Ganzhou Hospital-Nanfang Hospital, Southern Medical UniversityDepartment of Medical Imaging, Ganzhou Institute of Medical Imaging, Ganzhou Key Laboratory of Medical Imaging and Artificial Intelligence, Medical Imaging Center, Ganzhou People’s Hospital, The Affiliated Ganzhou Hospital of Nanchang University, Ganzhou Hospital-Nanfang Hospital, Southern Medical UniversityDepartment of Medical Imaging, Ganzhou Institute of Medical Imaging, Ganzhou Key Laboratory of Medical Imaging and Artificial Intelligence, Medical Imaging Center, Ganzhou People’s Hospital, The Affiliated Ganzhou Hospital of Nanchang University, Ganzhou Hospital-Nanfang Hospital, Southern Medical UniversityDepartment of Radiology, Shanghai Eastern Hepatobiliary Surgery Hospital, Second Military Medical UniversityAbstract Objective Intrahepatic cholangiocarcinoma (ICC) and hepatic inflammatory pseudotumours (IPTs) share similar imaging features, leading to unnecessary biopsies and surgeries. Accurate preoperative differentiation is essential. Current studies using traditional imaging analysis have limited accuracy. We developed a machine learning model based on clinical and CT radiomic features to improve diagnostic accuracy. Methods From May 2008 to January 2024, the data of 112 patients with ICC and 34 patients with hepatic IPTs who underwent preoperative plain and enhanced CT scans and whose diseases were confirmed by surgery and pathology were retrospectively analysed. A radiomic feature set, a clinical feature set, and a radiomic + clinical feature set were developed, and each was used to construct 14 machine learning models. The optimal hyperparameters were identified using fivefold cross-validation and a grid search. Finally, the area under the curve (AUC), accuracy, recall, precision, F1, Kappa value and other indicators were used to evaluate the performance of the models in the test sets to determine the optimal model for each feature subset. Results The machine learning model constructed with the radiomic features of all the CT sequences and the fused model constructed with both clinical features + all the CT sequence radiomic features performed well (AUC = 0.91 and 0.97, respectively), whereas the performance of the machine learning model constructed with the clinical features alone was relatively poor (AUC = 0.73). In terms of model performance in identifying the two diseases, the accuracy of the fused model was better in identifying ICCs than in identifying IPTs. Conclusion A diagnostic model constructed from clinical and CT radiomic features quickly differentiated between IPT from ICC. The model may be helpful for the preoperative identification of IPTs and ICC.https://doi.org/10.1186/s12885-025-14488-zIPTsICCCTRadiomicsMachine learning
spellingShingle Xiao-chun Wang
Jing-hong Liang
Xiao-yao Huang
Wen-jian Tang
Yan-mei He
Jun-yuan Zhong
Ling Zhang
Lun Lu
A machine learning model based on CT radiomics for preoperatively differentiating intrahepatic mass-type cholangiocarcinoma and inflammatory pseudotumours
BMC Cancer
IPTs
ICC
CT
Radiomics
Machine learning
title A machine learning model based on CT radiomics for preoperatively differentiating intrahepatic mass-type cholangiocarcinoma and inflammatory pseudotumours
title_full A machine learning model based on CT radiomics for preoperatively differentiating intrahepatic mass-type cholangiocarcinoma and inflammatory pseudotumours
title_fullStr A machine learning model based on CT radiomics for preoperatively differentiating intrahepatic mass-type cholangiocarcinoma and inflammatory pseudotumours
title_full_unstemmed A machine learning model based on CT radiomics for preoperatively differentiating intrahepatic mass-type cholangiocarcinoma and inflammatory pseudotumours
title_short A machine learning model based on CT radiomics for preoperatively differentiating intrahepatic mass-type cholangiocarcinoma and inflammatory pseudotumours
title_sort machine learning model based on ct radiomics for preoperatively differentiating intrahepatic mass type cholangiocarcinoma and inflammatory pseudotumours
topic IPTs
ICC
CT
Radiomics
Machine learning
url https://doi.org/10.1186/s12885-025-14488-z
work_keys_str_mv AT xiaochunwang amachinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT jinghongliang amachinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT xiaoyaohuang amachinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT wenjiantang amachinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT yanmeihe amachinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT junyuanzhong amachinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT lingzhang amachinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT lunlu amachinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT xiaochunwang machinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT jinghongliang machinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT xiaoyaohuang machinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT wenjiantang machinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT yanmeihe machinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT junyuanzhong machinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT lingzhang machinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours
AT lunlu machinelearningmodelbasedonctradiomicsforpreoperativelydifferentiatingintrahepaticmasstypecholangiocarcinomaandinflammatorypseudotumours