Deep Learning on Misaligned Dual-Energy Chest X-ray Images Using Paired Cycle-Consistent Generative Adversarial Networks

Ueda, Yasuyuki; Niu, Misato; Shimazaki, Riko; Yamazaki, Asumi; Seki, Masashi; Ishida, Takayuki

doi:https://doi.org/10.1007/s10278-025-01508-4

Deep Learning on Misaligned Dual-Energy Chest X-ray Images Using Paired Cycle-Consistent Generative Adversarial Networks

Ueda, Yasuyuki; Niu, Misato; Shimazaki, Riko et al.

Journal of Imaging Informatics in Medicine, 2025, 23(2)

Number of Access:524（2025-08-30 02:27 Counts）

Identifier to cite or link to this item: https://hdl.handle.net/11094/101115

Link to primary information

File	Format	Terms of use	Size	Views	Date.Available	Description	information
JDigitImagingInformmed_23_2_77	pdf	None	1.50 MB		2026.05.05

Item Information

Output File Export EndNote Basic Export Mendeley

Title	Deep Learning on Misaligned Dual-Energy Chest X-ray Images Using Paired Cycle-Consistent Generative Adversarial Networks
Creator	NRID 1000050780805 1000050780805 ORCID 0000-0003-0975-2164 0000-0003-0975-2164 researchmap permalink uedacci uedacci Scopus Author ID 37001274700 37001274700 CreatorName Ueda, Yasuyuki
	CreatorName Niu, Misato
	CreatorName Shimazaki, Riko
	NRID 1000080939588 1000080939588 researchmap permalink asumi_yamazaki asumi_yamazaki Scopus Author ID 16551193100 16551193100 CreatorName Yamazaki, Asumi
	CreatorName Seki, Masashi
	NRID 1000060309656 1000060309656 ORCID 0000-0002-6851-0598 0000-0002-6851-0598 researchmap permalink Taka_Ishida00 Taka_Ishida00 Scopus Author ID 35406489400 35406489400 CreatorName Ishida, Takayuki
Subject	Dual-energy subtraction
	Chest X-ray
	Motion artifacts
	Generative adversarial networks
	Cycle-consistent generative adversarial networks
	Deep learning
Description	Ueda Yasuyuki, Niu Misato, Shimazaki Riko, et al. Deep Learning on Misaligned Dual-Energy Chest X-ray Images Using Paired Cycle-Consistent Generative Adversarial Networks. Journal of Imaging Informatics in Medicine 23, 77 (2025); https://doi.org/10.1007/s10278-025-01508-4.
Description	This version of the article has been accepted for publication, after peer review (when applicable) and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: https://doi.org/10.1007/s10278-025-01508-4.
Abstract	Dual-energy subtraction (DES) chest X-ray images (CXRs) are often affected by motion artifacts resulting from patients’ voluntary or involuntary movements, even in clinical settings. Additionally, the mediastinum and upper abdominal regions in low-energy (LE) CXRs are susceptible to signal insufficiency due to inadequate input photon numbers. Current image processing techniques for removing motion artifacts and statistical noise from DES-CXRs are insufficient, and potential algorithms for these tasks remain largely unexplored. We propose a framework based on paired cycle-consistency adversarial generative networks to effectively remove motion artifacts and statistical noise from DES-CXRs. The proposed method incorporates ensemble discriminators, differentiable augmentation, anti-aliased convolution layers, and a basic 8-layer U-Net generator. This method was trained and tested using a clinical image dataset comprising data of 600 examinations of individuals who underwent dual-energy chest X-ray imaging for diagnostic purposes, using a sixfold cross-validation approach. It demonstrated a remarkable improvement in motion artifact suppression in terms of an analysis of full width at the 10-percent maximum improved from 0.216 ± 0.0720 to 0.200 ± 0.0783 for the left lung region of interests including the cardiac region. Furthermore, it outperformed the method in a previous study in terms of a peak signal-to-noise ratio of 50.7 ± 3.68, structural similarity index of 0.997 ± 0.0152 for LE images, and Fréchet inception distance of 85.0 ± 3.52 for bone-suppressed DES images. The proposed method significantly outperforms existing techniques for removing motion artifacts and statistical noise and shows strong potential for clinical applications in chest X-ray imaging.
Publisher	Springer Science and Business Media LLC
Source Title	Journal of Imaging Informatics in Medicine
Volume (Issue)	23(2)
Date of Issued	2025-05-05
Language	English
Handle URL	https://hdl.handle.net/11094/101115
eISSN	29482933
Relation.isVersionOf	DOI (Publisher Version) https://doi.org/10.1007/s10278-025-01508-4
Access Rights	embargoed access
CopyRight	The full-text file will be made open to the public on 5 May 2026 in accordance with the publisher's policy.
oaire:version	Accepted Manuscript
Category	学術雑誌論文 Journal Article

Resource Type	journal article
Local Type	Journal Article
DCMI Type	Text
DCTERMS.bibliographicCitation	Journal of Imaging Informatics in Medicine. 2025, 23(2)
DC.title	Deep Learning on Misaligned Dual-Energy Chest X-ray Images Using Paired Cycle-Consistent Generative Adversarial Networks
DC.creator	Ueda, Yasuyuki
	Niu, Misato
	Shimazaki, Riko
	Yamazaki, Asumi
	Seki, Masashi
	Ishida, Takayuki
DC.publisher	Springer Science and Business Media LLC
DC.language' scheme='DCTERMS.RFC1766	English
DC.type' scheme='DCTERMS.DCMIType	Text
DCTERMS.issued' scheme='DCTERMS.W3CDTF	2025-05-05
DC.identifier	https://doi.org/10.1007/s10278-025-01508-4
DC.subject	Dual-energy subtraction
	Chest X-ray
	Motion artifacts
	Generative adversarial networks
	Cycle-consistent generative adversarial networks
	Deep learning
DC.description	Ueda Yasuyuki, Niu Misato, Shimazaki Riko, et al. Deep Learning on Misaligned Dual-Energy Chest X-ray Images Using Paired Cycle-Consistent Generative Adversarial Networks. Journal of Imaging Informatics in Medicine 23, 77 (2025); https://doi.org/10.1007/s10278-025-01508-4.
DC.description	This version of the article has been accepted for publication, after peer review (when applicable) and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: https://doi.org/10.1007/s10278-025-01508-4.
DCTERMS.abstract	Dual-energy subtraction (DES) chest X-ray images (CXRs) are often affected by motion artifacts resulting from patients’ voluntary or involuntary movements, even in clinical settings. Additionally, the mediastinum and upper abdominal regions in low-energy (LE) CXRs are susceptible to signal insufficiency due to inadequate input photon numbers. Current image processing techniques for removing motion artifacts and statistical noise from DES-CXRs are insufficient, and potential algorithms for these tasks remain largely unexplored. We propose a framework based on paired cycle-consistency adversarial generative networks to effectively remove motion artifacts and statistical noise from DES-CXRs. The proposed method incorporates ensemble discriminators, differentiable augmentation, anti-aliased convolution layers, and a basic 8-layer U-Net generator. This method was trained and tested using a clinical image dataset comprising data of 600 examinations of individuals who underwent dual-energy chest X-ray imaging for diagnostic purposes, using a sixfold cross-validation approach. It demonstrated a remarkable improvement in motion artifact suppression in terms of an analysis of full width at the 10-percent maximum improved from 0.216 ± 0.0720 to 0.200 ± 0.0783 for the left lung region of interests including the cardiac region. Furthermore, it outperformed the method in a previous study in terms of a peak signal-to-noise ratio of 50.7 ± 3.68, structural similarity index of 0.997 ± 0.0152 for LE images, and Fréchet inception distance of 85.0 ± 3.52 for bone-suppressed DES images. The proposed method significantly outperforms existing techniques for removing motion artifacts and statistical noise and shows strong potential for clinical applications in chest X-ray imaging.
DC.rights	The full-text file will be made open to the public on 5 May 2026 in accordance with the publisher's policy.
DC.format	101115
citation_title	Deep Learning on Misaligned Dual-Energy Chest X-ray Images Using Paired Cycle-Consistent Generative Adversarial Networks
citation_author	Ueda, Yasuyuki
	Niu, Misato
	Shimazaki, Riko
	Yamazaki, Asumi
	Seki, Masashi
	Ishida, Takayuki
citation_publisher	Springer Science and Business Media LLC
citation_language	English
citation_date	2025-05-05
citation_journal_title	Journal of Imaging Informatics in Medicine
citation_volume	23
citation_issue	2
citation_doi	https://doi.org/10.1007/s10278-025-01508-4
citation_public_url	https://hdl.handle.net/11094/101115

Deep Learning on Misaligned Dual-Energy Chest X-ray Images Using Paired Cycle-Consistent Generative Adversarial Networks

Link to primary information

Item Information

Show details