Interpretable and generalizable deep learning model for preoperative assessment of microvascular invasion and outcome in hepatocellular carcinoma based on MRI: a multicenter study

Abstract Objectives This study aimed to develop an interpretable, domain-generalizable deep learning model for microvascular invasion (MVI) assessment in hepatocellular carcinoma (HCC). Methods Utilizing a retrospective dataset of 546 HCC patients from five centers, we developed and validated a clin...

Full description

Saved in:
Bibliographic Details
Main Authors: Xue Dong, Xibin Jia, Wei Zhang, Jingxuan Zhang, Hui Xu, Lixue Xu, Caili Ma, Hongjie Hu, Jiawen Luo, Jingfeng Zhang, Zhenchang Wang, Wenbin Ji, Dawei Yang, Zhenghan Yang
Format: Article
Language:English
Published: SpringerOpen 2025-07-01
Series:Insights into Imaging
Subjects:
Online Access:https://doi.org/10.1186/s13244-025-02035-0
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849399574344499200
author Xue Dong
Xibin Jia
Wei Zhang
Jingxuan Zhang
Hui Xu
Lixue Xu
Caili Ma
Hongjie Hu
Jiawen Luo
Jingfeng Zhang
Zhenchang Wang
Wenbin Ji
Dawei Yang
Zhenghan Yang
author_facet Xue Dong
Xibin Jia
Wei Zhang
Jingxuan Zhang
Hui Xu
Lixue Xu
Caili Ma
Hongjie Hu
Jiawen Luo
Jingfeng Zhang
Zhenchang Wang
Wenbin Ji
Dawei Yang
Zhenghan Yang
author_sort Xue Dong
collection DOAJ
description Abstract Objectives This study aimed to develop an interpretable, domain-generalizable deep learning model for microvascular invasion (MVI) assessment in hepatocellular carcinoma (HCC). Methods Utilizing a retrospective dataset of 546 HCC patients from five centers, we developed and validated a clinical-radiological model and deep learning models aimed at MVI prediction. The models were developed on a dataset of 263 cases consisting of data from three centers, internally validated on a set of 66 patients, and externally tested on two independent sets. An adversarial network-based deep learning (AD-DL) model was developed to learn domain-invariant features from multiple centers within the training set. The area under the receiver operating characteristic curve (AUC) was calculated using pathological MVI status. With the best-performed model, early recurrence-free survival (ERFS) stratification was validated on the external test set by the log-rank test, and the differentially expressed genes (DEGs) associated with MVI status were tested on the RNA sequencing analysis of the Cancer Imaging Archive. Results The AD-DL model demonstrated the highest diagnostic performance and generalizability with an AUC of 0.793 in the internal test set, 0.801 in external test set 1, and 0.773 in external test set 2. The model’s prediction of MVI status also demonstrated a significant correlation with ERFS (p = 0.048). DEGs associated with MVI status were primarily enriched in the metabolic processes and the Wnt signaling pathway, and the epithelial-mesenchymal transition process. Conclusions The AD-DL model allows preoperative MVI prediction and ERFS stratification in HCC patients, which has a good generalizability and biological interpretability. Critical relevance statement The adversarial network-based deep learning model predicts MVI status well in HCC patients and demonstrates good generalizability. By integrating bioinformatics analysis of the model’s predictions, it achieves biological interpretability, facilitating its clinical translation. Key Points Current MVI assessment models for HCC lack interpretability and generalizability. The adversarial network-based model's performance surpassed clinical radiology and squeeze-and-excitation network-based models. Biological function analysis was employed to enhance the interpretability and clinical translatability of the adversarial network-based model. Graphical Abstract
format Article
id doaj-art-bc02b53d99ab48b7b764b02c02085a4a
institution Kabale University
issn 1869-4101
language English
publishDate 2025-07-01
publisher SpringerOpen
record_format Article
series Insights into Imaging
spelling doaj-art-bc02b53d99ab48b7b764b02c02085a4a2025-08-20T03:38:18ZengSpringerOpenInsights into Imaging1869-41012025-07-0116111310.1186/s13244-025-02035-0Interpretable and generalizable deep learning model for preoperative assessment of microvascular invasion and outcome in hepatocellular carcinoma based on MRI: a multicenter studyXue Dong0Xibin Jia1Wei Zhang2Jingxuan Zhang3Hui Xu4Lixue Xu5Caili Ma6Hongjie Hu7Jiawen Luo8Jingfeng Zhang9Zhenchang Wang10Wenbin Ji11Dawei Yang12Zhenghan Yang13Department of Radiology, Beijing Friendship Hospital, Capital Medical UniversityCollege of Computer Science, Beijing University of TechnologyCollege of Computer Science, Beijing University of TechnologyDepartment of Radiology, Beijing Friendship Hospital, Capital Medical UniversityDepartment of Radiology, Beijing Friendship Hospital, Capital Medical UniversityDepartment of Radiology, Beijing Friendship Hospital, Capital Medical UniversityDepartment of Radiology, Beijing Longfu HospitalDepartment of Radiology, Sir Run Run Shaw Hospital, Zhejiang University School of MedicineDepartment of Radiology, The Second Hospital of Dalian Medical UniversityDepartment of Radiology, Ningbo No. 2 HospitalDepartment of Radiology, Beijing Friendship Hospital, Capital Medical UniversityDepartment of Radiology, Taizhou Hospital of Zhejiang Province Affiliated to Wenzhou Medical UniversityDepartment of Radiology, Beijing Friendship Hospital, Capital Medical UniversityDepartment of Radiology, Beijing Friendship Hospital, Capital Medical UniversityAbstract Objectives This study aimed to develop an interpretable, domain-generalizable deep learning model for microvascular invasion (MVI) assessment in hepatocellular carcinoma (HCC). Methods Utilizing a retrospective dataset of 546 HCC patients from five centers, we developed and validated a clinical-radiological model and deep learning models aimed at MVI prediction. The models were developed on a dataset of 263 cases consisting of data from three centers, internally validated on a set of 66 patients, and externally tested on two independent sets. An adversarial network-based deep learning (AD-DL) model was developed to learn domain-invariant features from multiple centers within the training set. The area under the receiver operating characteristic curve (AUC) was calculated using pathological MVI status. With the best-performed model, early recurrence-free survival (ERFS) stratification was validated on the external test set by the log-rank test, and the differentially expressed genes (DEGs) associated with MVI status were tested on the RNA sequencing analysis of the Cancer Imaging Archive. Results The AD-DL model demonstrated the highest diagnostic performance and generalizability with an AUC of 0.793 in the internal test set, 0.801 in external test set 1, and 0.773 in external test set 2. The model’s prediction of MVI status also demonstrated a significant correlation with ERFS (p = 0.048). DEGs associated with MVI status were primarily enriched in the metabolic processes and the Wnt signaling pathway, and the epithelial-mesenchymal transition process. Conclusions The AD-DL model allows preoperative MVI prediction and ERFS stratification in HCC patients, which has a good generalizability and biological interpretability. Critical relevance statement The adversarial network-based deep learning model predicts MVI status well in HCC patients and demonstrates good generalizability. By integrating bioinformatics analysis of the model’s predictions, it achieves biological interpretability, facilitating its clinical translation. Key Points Current MVI assessment models for HCC lack interpretability and generalizability. The adversarial network-based model's performance surpassed clinical radiology and squeeze-and-excitation network-based models. Biological function analysis was employed to enhance the interpretability and clinical translatability of the adversarial network-based model. Graphical Abstracthttps://doi.org/10.1186/s13244-025-02035-0Hepatocellular carcinomaMicrovascular invasionDeep learning modelGeneralizationInterpretability
spellingShingle Xue Dong
Xibin Jia
Wei Zhang
Jingxuan Zhang
Hui Xu
Lixue Xu
Caili Ma
Hongjie Hu
Jiawen Luo
Jingfeng Zhang
Zhenchang Wang
Wenbin Ji
Dawei Yang
Zhenghan Yang
Interpretable and generalizable deep learning model for preoperative assessment of microvascular invasion and outcome in hepatocellular carcinoma based on MRI: a multicenter study
Insights into Imaging
Hepatocellular carcinoma
Microvascular invasion
Deep learning model
Generalization
Interpretability
title Interpretable and generalizable deep learning model for preoperative assessment of microvascular invasion and outcome in hepatocellular carcinoma based on MRI: a multicenter study
title_full Interpretable and generalizable deep learning model for preoperative assessment of microvascular invasion and outcome in hepatocellular carcinoma based on MRI: a multicenter study
title_fullStr Interpretable and generalizable deep learning model for preoperative assessment of microvascular invasion and outcome in hepatocellular carcinoma based on MRI: a multicenter study
title_full_unstemmed Interpretable and generalizable deep learning model for preoperative assessment of microvascular invasion and outcome in hepatocellular carcinoma based on MRI: a multicenter study
title_short Interpretable and generalizable deep learning model for preoperative assessment of microvascular invasion and outcome in hepatocellular carcinoma based on MRI: a multicenter study
title_sort interpretable and generalizable deep learning model for preoperative assessment of microvascular invasion and outcome in hepatocellular carcinoma based on mri a multicenter study
topic Hepatocellular carcinoma
Microvascular invasion
Deep learning model
Generalization
Interpretability
url https://doi.org/10.1186/s13244-025-02035-0
work_keys_str_mv AT xuedong interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT xibinjia interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT weizhang interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT jingxuanzhang interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT huixu interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT lixuexu interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT cailima interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT hongjiehu interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT jiawenluo interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT jingfengzhang interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT zhenchangwang interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT wenbinji interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT daweiyang interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy
AT zhenghanyang interpretableandgeneralizabledeeplearningmodelforpreoperativeassessmentofmicrovascularinvasionandoutcomeinhepatocellularcarcinomabasedonmriamulticenterstudy