Domain Shift Analysis in Chest Radiographs Classification in a Veterans Healthcare Administration Population

Mayanka Chandrashekar; Ian Goethert; Md Inzamam Ul Haque; Benjamin McMahon; Sayera Dhaubhadel; Kathryn Knight; Joseph Erdos; Donna Reagan; Caroline Taylor; Peter Kuzmak; John Michael Gaziano; Eileen McAllister; Lauren Costa; Yuk Lam Ho; Kelly Cho; Suzanne Tamang; Samah Fodeh-Jarad; Olga S. Ovchinnikova; Amy C. Justice; Jacob Hinkle; Ioana Danciu

doi:10.1007/s10278-025-01494-7

Domain Shift Analysis in Chest Radiographs Classification in a Veterans Healthcare Administration Population

Mayanka Chandrashekar
, Ian Goethert
, Md Inzamam Ul Haque
, Benjamin McMahon
, Sayera Dhaubhadel
, Kathryn Knight
, Joseph Erdos
, Donna Reagan
, Caroline Taylor
, Peter Kuzmak
, John Michael Gaziano
, Eileen McAllister
, Lauren Costa
, Yuk Lam Ho
, Kelly Cho
, Suzanne Tamang
, Samah Fodeh-Jarad
, Olga S. Ovchinnikova
, Amy C. Justice
, Jacob Hinkle

Ioana Danciu

Research output: Contribution to journal › Article › peer-review

Abstract

This study aims to assess the impact of domain shift on chest X-ray classification accuracy and to analyze the influence of ground truth label quality and demographic factors such as age group, sex, and study year. We used a DenseNet121 model pre-trained MIMIC-CXR dataset for deep learning-based multi-label classification using ground truth labels from radiology reports extracted using the CheXpert and CheXbert Labeler. We compared the performance of the 14 chest X-ray labels on the MIMIC-CXR and Veterans Healthcare Administration chest X-ray dataset (VA-CXR). The validation of ground truth and the assessment of multi-label classification performance across various NLP extraction tools revealed that the VA-CXR dataset exhibited lower disagreement rates than the MIMIC-CXR datasets. Additionally, there were notable differences in AUC scores between models utilizing CheXpert and CheXbert. When evaluating multi-label classification performance across different datasets, minimal domain shift was observed in the unseen VA dataset, except for the label “Enlarged Cardiomediastinum.” The subgroup with the most significant variations in multi-label classification performance was study year. These findings underscore the importance of considering domain shift in chest X-ray classification tasks, paying particular attention to the temporality of the exam. Our study reveals the significant impact of domain shift and demographic factors on chest X-ray classification, emphasizing the need for improved transfer learning and robust model development. Addressing these challenges is crucial for advancing medical imaging research and improving patient care.

Original language	English
Pages (from-to)	484-499
Number of pages	16
Journal	Journal of Imaging Informatics in Medicine
Volume	39
Issue number	1
DOIs	https://doi.org/10.1007/s10278-025-01494-7
State	Published - Feb 2026

Funding

This work is sponsored by the US Department of Veterans Affairs using resources from the Knowledge Discovery Infrastructure, which is located at the Oak Ridge National Laboratory, and supported by the Office of Science of the U.S. Department of Energy. This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US government retains, and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript or allow others to do so for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan). Notice: Office of Science of the U.S. Department of Energy. This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US government retains, and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan ( http://energy.gov/downloads/doe-public-access-plan ).

Keywords

Chest X-ray image classification
Domain shift
Multi-label classification

Access to Document

10.1007/s10278-025-01494-7

Cite this

Chandrashekar, M., Goethert, I., Haque, M. I. U., McMahon, B., Dhaubhadel, S., Knight, K., Erdos, J., Reagan, D., Taylor, C., Kuzmak, P., Gaziano, J. M., McAllister, E., Costa, L., Ho, Y. L., Cho, K., Tamang, S., Fodeh-Jarad, S., Ovchinnikova, O. S., Justice, A. C., ... Danciu, I. (2026). Domain Shift Analysis in Chest Radiographs Classification in a Veterans Healthcare Administration Population. Journal of Imaging Informatics in Medicine, 39(1), 484-499. https://doi.org/10.1007/s10278-025-01494-7

@article{1254b65d74ce4c4bbbc3b8d6eb3d675d,

title = "Domain Shift Analysis in Chest Radiographs Classification in a Veterans Healthcare Administration Population",

abstract = "This study aims to assess the impact of domain shift on chest X-ray classification accuracy and to analyze the influence of ground truth label quality and demographic factors such as age group, sex, and study year. We used a DenseNet121 model pre-trained MIMIC-CXR dataset for deep learning-based multi-label classification using ground truth labels from radiology reports extracted using the CheXpert and CheXbert Labeler. We compared the performance of the 14 chest X-ray labels on the MIMIC-CXR and Veterans Healthcare Administration chest X-ray dataset (VA-CXR). The validation of ground truth and the assessment of multi-label classification performance across various NLP extraction tools revealed that the VA-CXR dataset exhibited lower disagreement rates than the MIMIC-CXR datasets. Additionally, there were notable differences in AUC scores between models utilizing CheXpert and CheXbert. When evaluating multi-label classification performance across different datasets, minimal domain shift was observed in the unseen VA dataset, except for the label “Enlarged Cardiomediastinum.” The subgroup with the most significant variations in multi-label classification performance was study year. These findings underscore the importance of considering domain shift in chest X-ray classification tasks, paying particular attention to the temporality of the exam. Our study reveals the significant impact of domain shift and demographic factors on chest X-ray classification, emphasizing the need for improved transfer learning and robust model development. Addressing these challenges is crucial for advancing medical imaging research and improving patient care.",

keywords = "Chest X-ray image classification, Domain shift, Multi-label classification",

author = "Mayanka Chandrashekar and Ian Goethert and Haque, \{Md Inzamam Ul\} and Benjamin McMahon and Sayera Dhaubhadel and Kathryn Knight and Joseph Erdos and Donna Reagan and Caroline Taylor and Peter Kuzmak and Gaziano, \{John Michael\} and Eileen McAllister and Lauren Costa and Ho, \{Yuk Lam\} and Kelly Cho and Suzanne Tamang and Samah Fodeh-Jarad and Ovchinnikova, \{Olga S.\} and Justice, \{Amy C.\} and Jacob Hinkle and Ioana Danciu",

note = "Publisher Copyright: {\textcopyright} The Author(s) 2025.",

year = "2026",

month = feb,

doi = "10.1007/s10278-025-01494-7",

language = "English",

volume = "39",

pages = "484--499",

journal = "Journal of Imaging Informatics in Medicine",

issn = "2948-2933",

publisher = "Springer",

number = "1",

}

Chandrashekar, M , Goethert, I, Haque, MIU, McMahon, B, Dhaubhadel, S, Knight, K, Erdos, J, Reagan, D, Taylor, C, Kuzmak, P, Gaziano, JM, McAllister, E, Costa, L, Ho, YL, Cho, K, Tamang, S, Fodeh-Jarad, S, Ovchinnikova, OS, Justice, AC, Hinkle, J & Danciu, I 2026, 'Domain Shift Analysis in Chest Radiographs Classification in a Veterans Healthcare Administration Population', Journal of Imaging Informatics in Medicine, vol. 39, no. 1, pp. 484-499. https://doi.org/10.1007/s10278-025-01494-7

TY - JOUR

T1 - Domain Shift Analysis in Chest Radiographs Classification in a Veterans Healthcare Administration Population

AU - Chandrashekar, Mayanka

AU - Goethert, Ian

AU - Haque, Md Inzamam Ul

AU - McMahon, Benjamin

AU - Dhaubhadel, Sayera

AU - Knight, Kathryn

AU - Erdos, Joseph

AU - Reagan, Donna

AU - Taylor, Caroline

AU - Kuzmak, Peter

AU - Gaziano, John Michael

AU - McAllister, Eileen

AU - Costa, Lauren

AU - Ho, Yuk Lam

AU - Cho, Kelly

AU - Tamang, Suzanne

AU - Fodeh-Jarad, Samah

AU - Ovchinnikova, Olga S.

AU - Justice, Amy C.

AU - Hinkle, Jacob

AU - Danciu, Ioana

PY - 2026/2

Y1 - 2026/2

N2 - This study aims to assess the impact of domain shift on chest X-ray classification accuracy and to analyze the influence of ground truth label quality and demographic factors such as age group, sex, and study year. We used a DenseNet121 model pre-trained MIMIC-CXR dataset for deep learning-based multi-label classification using ground truth labels from radiology reports extracted using the CheXpert and CheXbert Labeler. We compared the performance of the 14 chest X-ray labels on the MIMIC-CXR and Veterans Healthcare Administration chest X-ray dataset (VA-CXR). The validation of ground truth and the assessment of multi-label classification performance across various NLP extraction tools revealed that the VA-CXR dataset exhibited lower disagreement rates than the MIMIC-CXR datasets. Additionally, there were notable differences in AUC scores between models utilizing CheXpert and CheXbert. When evaluating multi-label classification performance across different datasets, minimal domain shift was observed in the unseen VA dataset, except for the label “Enlarged Cardiomediastinum.” The subgroup with the most significant variations in multi-label classification performance was study year. These findings underscore the importance of considering domain shift in chest X-ray classification tasks, paying particular attention to the temporality of the exam. Our study reveals the significant impact of domain shift and demographic factors on chest X-ray classification, emphasizing the need for improved transfer learning and robust model development. Addressing these challenges is crucial for advancing medical imaging research and improving patient care.

AB - This study aims to assess the impact of domain shift on chest X-ray classification accuracy and to analyze the influence of ground truth label quality and demographic factors such as age group, sex, and study year. We used a DenseNet121 model pre-trained MIMIC-CXR dataset for deep learning-based multi-label classification using ground truth labels from radiology reports extracted using the CheXpert and CheXbert Labeler. We compared the performance of the 14 chest X-ray labels on the MIMIC-CXR and Veterans Healthcare Administration chest X-ray dataset (VA-CXR). The validation of ground truth and the assessment of multi-label classification performance across various NLP extraction tools revealed that the VA-CXR dataset exhibited lower disagreement rates than the MIMIC-CXR datasets. Additionally, there were notable differences in AUC scores between models utilizing CheXpert and CheXbert. When evaluating multi-label classification performance across different datasets, minimal domain shift was observed in the unseen VA dataset, except for the label “Enlarged Cardiomediastinum.” The subgroup with the most significant variations in multi-label classification performance was study year. These findings underscore the importance of considering domain shift in chest X-ray classification tasks, paying particular attention to the temporality of the exam. Our study reveals the significant impact of domain shift and demographic factors on chest X-ray classification, emphasizing the need for improved transfer learning and robust model development. Addressing these challenges is crucial for advancing medical imaging research and improving patient care.

KW - Chest X-ray image classification

KW - Domain shift

KW - Multi-label classification

UR - https://www.scopus.com/pages/publications/105007709862

U2 - 10.1007/s10278-025-01494-7

DO - 10.1007/s10278-025-01494-7

M3 - Article

AN - SCOPUS:105007709862

SN - 2948-2933

VL - 39

SP - 484

EP - 499

JO - Journal of Imaging Informatics in Medicine

JF - Journal of Imaging Informatics in Medicine

IS - 1

ER -

Domain Shift Analysis in Chest Radiographs Classification in a Veterans Healthcare Administration Population

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this