EUVIMEDEUVIMED entdecken
Open Access

Impact of Exposure Parameters on Deep Learning Models in Chest Radiography and Implications for Deployment

Han-Jay Shu iD, Shun-Ting Chang iD, Rodrigo Rosa Gameiro iD, Judy Wawira Gichoya iD, Leo Anthony Celi, Po-Chih Kuo iD

Radiology: Artificial Intelligence · 01.07.2026 · 8 (4)

DOI: 10.1148/ryai.250731 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Radiology: Artificial Intelligence, erschienen am 01.07.2026. Verfasst von Han-Jay Shu und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Computer Science, National Tsing Hua University, Hsinchu, Taiwan. Hauptthemen laut MEDLINE: Deep Learning, Radiography, Thoracic, COVID-19, Pneumothorax. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 28 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: National Science and Technology Council, Taiwan, National Heart, Lung, and Blood Institute, National Institutes of Health.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

An algorithmic framework was developed to identify and quantify shortcut learning and bias driven by exposure parameters in chest radiographs, revealing hidden sources of bias in medical artificial intelligence.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningRadiography, ThoracicCOVID-19PneumothoraxRadiographic Image Interpretation, Computer-AssistedHumansRetrospective StudiesFemaleMiddle AgedMaleSARS-CoV-2Radiation DosageAged

Stichwörter der Autor:innen

Chest RadiographComputer Aided Diagnosis (CAD)Convolutional Neural Network (CNN)Deep LearningDiagnosisExposure ParametersFairnessFeature DetectionLungShortcut Learning

Autor:innen und Institutionen

  • Department of Computer Science, National Tsing Hua University, Hsinchu, Taiwan.
  • Laboratory for Computational Physiology, Massachusetts Institute of Technology, Cambridge, Mass.
  • Division of Pulmonary Critical Care and Sleep Medicine, Beth Israel Deaconess Medical Center, Boston, Mass.
  • Department of Radiology, Emory University, Atlanta, Ga.
  • Department of Biostatistics, Harvard T. H. Chan School of Public Health, Boston, Mass.

ORCID-Profile: Han-Jay Shu, Shun-Ting Chang, Rodrigo Rosa Gameiro, Judy Wawira Gichoya, Po-Chih Kuo

Förderung

  • National Science and Technology Council, Taiwan
  • National Heart, Lung, and Blood Institute
  • National Institutes of Health
  • Health Discovery and Innovation in Africa and Bridge2AI
  • National Science Foundation through the Innovative Technology Experiences for Students and Teachers

Zitieren

APA

Shu, H. J., Chang, S. T., Gameiro, R. R., Gichoya, J. W., Celi, L. A., & Kuo, P. C. (2026). Impact of Exposure Parameters on Deep Learning Models in Chest Radiography and Implications for Deployment. Radiology: Artificial Intelligence, 8(4). https://doi.org/10.1148/ryai.250731

Vancouver

Shu HJ, Chang ST, Gameiro RR, Gichoya JW, Celi LA, Kuo PC. Impact of Exposure Parameters on Deep Learning Models in Chest Radiography and Implications for Deployment. Radiology: Artificial Intelligence. 2026;8(4). doi:10.1148/ryai.250731

BibTeX
@article{Shu2026,
  author = {Shu, Han-Jay and Chang, Shun-Ting and Gameiro, Rodrigo Rosa and Gichoya, Judy Wawira and Celi, Leo Anthony and Kuo, Po-Chih},
  title = {Impact of Exposure Parameters on Deep Learning Models in Chest Radiography and Implications for Deployment},
  journal = {Radiology: Artificial Intelligence},
  year = {2026},
  volume = {8},
  number = {4},
  doi = {10.1148/ryai.250731}
}
Kostenlos starten →