A modified generative adversarial networks method for assisting the diagnosis of deep venous thrombosis complications in stroke patients

Abstract Discriminate deep vein thrombosis, one of the complications in early stroke patients, in order to assist in diagnosis. We have constructed a new method called ACWGAN by combining ACGAN and WGAN methods for data augmentation to to enhance the data of stroke early rehabilitation patients admi...

Full description

Saved in:
Bibliographic Details
Main Authors: Ying Wang, Kaipeng Wang, Jinjun Hou, Ziqi Ye, Changsheng Lin, Xueping Li
Format: Article
Language:English
Published: Nature Portfolio 2025-07-01
Series:Scientific Reports
Online Access:https://doi.org/10.1038/s41598-025-04880-x
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849769163714723840
author Ying Wang
Kaipeng Wang
Jinjun Hou
Ziqi Ye
Changsheng Lin
Xueping Li
author_facet Ying Wang
Kaipeng Wang
Jinjun Hou
Ziqi Ye
Changsheng Lin
Xueping Li
author_sort Ying Wang
collection DOAJ
description Abstract Discriminate deep vein thrombosis, one of the complications in early stroke patients, in order to assist in diagnosis. We have constructed a new method called ACWGAN by combining ACGAN and WGAN methods for data augmentation to to enhance the data of stroke early rehabilitation patients admitted to Nanjing First Hospital from 2017 to 2021, followed by analysis of complications, and compared it with 20 other commonly used data augmentation methods. A total of 7110 patients were included in the analysis of this study, the AUC value of the discriminative model ranges from 0.688 to 0.805 after data augmentation using traditional machine learning methods. When deep learning methods (GAN-based methods) are employed, the AUC value can surpass 0.866. Our proposed ACWGAN method maintains efficiency comparable to GAN, ACGAN, and WGAN, while achieving an even higher AUC value for the model trained on the augmented dataset, exceeding 0.938. The ACWGAN method effectively improves the diversity and accuracy of the model while ensuring the stability of the network, indicating that it can accurately assist in diagnosing the prevalence of DVT in patients.
format Article
id doaj-art-f0c59559e2384d2eab8cd26b1888be7c
institution DOAJ
issn 2045-2322
language English
publishDate 2025-07-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-f0c59559e2384d2eab8cd26b1888be7c2025-08-20T03:03:32ZengNature PortfolioScientific Reports2045-23222025-07-0115111310.1038/s41598-025-04880-xA modified generative adversarial networks method for assisting the diagnosis of deep venous thrombosis complications in stroke patientsYing Wang0Kaipeng Wang1Jinjun Hou2Ziqi Ye3Changsheng Lin4Xueping Li5School of Mathematics and Statistics, Nanjing University of Science and TechnologySchool of Mathematics and Statistics, Nanjing University of Science and TechnologyDepartment of Rehabilitation Medicine, Nanjing First Hospital, Nanjing Medical UniversityDepartment of Rehabilitation Medicine, Nanjing First Hospital, Nanjing Medical UniversityDepartment of Rehabilitation Medicine, Nanjing First Hospital, Nanjing Medical UniversityDepartment of Rehabilitation Medicine, Nanjing First Hospital, Nanjing Medical UniversityAbstract Discriminate deep vein thrombosis, one of the complications in early stroke patients, in order to assist in diagnosis. We have constructed a new method called ACWGAN by combining ACGAN and WGAN methods for data augmentation to to enhance the data of stroke early rehabilitation patients admitted to Nanjing First Hospital from 2017 to 2021, followed by analysis of complications, and compared it with 20 other commonly used data augmentation methods. A total of 7110 patients were included in the analysis of this study, the AUC value of the discriminative model ranges from 0.688 to 0.805 after data augmentation using traditional machine learning methods. When deep learning methods (GAN-based methods) are employed, the AUC value can surpass 0.866. Our proposed ACWGAN method maintains efficiency comparable to GAN, ACGAN, and WGAN, while achieving an even higher AUC value for the model trained on the augmented dataset, exceeding 0.938. The ACWGAN method effectively improves the diversity and accuracy of the model while ensuring the stability of the network, indicating that it can accurately assist in diagnosing the prevalence of DVT in patients.https://doi.org/10.1038/s41598-025-04880-x
spellingShingle Ying Wang
Kaipeng Wang
Jinjun Hou
Ziqi Ye
Changsheng Lin
Xueping Li
A modified generative adversarial networks method for assisting the diagnosis of deep venous thrombosis complications in stroke patients
Scientific Reports
title A modified generative adversarial networks method for assisting the diagnosis of deep venous thrombosis complications in stroke patients
title_full A modified generative adversarial networks method for assisting the diagnosis of deep venous thrombosis complications in stroke patients
title_fullStr A modified generative adversarial networks method for assisting the diagnosis of deep venous thrombosis complications in stroke patients
title_full_unstemmed A modified generative adversarial networks method for assisting the diagnosis of deep venous thrombosis complications in stroke patients
title_short A modified generative adversarial networks method for assisting the diagnosis of deep venous thrombosis complications in stroke patients
title_sort modified generative adversarial networks method for assisting the diagnosis of deep venous thrombosis complications in stroke patients
url https://doi.org/10.1038/s41598-025-04880-x
work_keys_str_mv AT yingwang amodifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients
AT kaipengwang amodifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients
AT jinjunhou amodifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients
AT ziqiye amodifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients
AT changshenglin amodifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients
AT xuepingli amodifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients
AT yingwang modifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients
AT kaipengwang modifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients
AT jinjunhou modifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients
AT ziqiye modifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients
AT changshenglin modifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients
AT xuepingli modifiedgenerativeadversarialnetworksmethodforassistingthediagnosisofdeepvenousthrombosiscomplicationsinstrokepatients