Evaluating the robustness of deep learning models trained to diagnose idiopathic pulmonary fibrosis using a retrospective study
Preliminary findings demonstrated the lack of robustness of IPF diagnosis model when the DL-based model is applied to CT series collected under different imaging protocols, which indicated that care should be taken as to the acquisition and reconstruction conditions used when developing and deploying DL models into clinical practice.
Authors
- Jonathan Goldin (ORCID: https://orcid.org/0000-0003-3007-9947)
- Jin Woo Song (ORCID: https://orcid.org/0000-0001-5121-3522)
- Grace Hyun J. Kim (ORCID: https://orcid.org/0000-0003-1225-3489)
- Michael F. McNitt‐Gray (ORCID: https://orcid.org/0000-0003-3004-4613)
- Wenxi Yu (ORCID: https://orcid.org/0000-0003-4893-9618)
Institutions
- Ulsan College (KR)
- University of California, Los Angeles (US)
- Asan Medical Center (KR)
- University of Ulsan (KR)
Publication Details
- Journal
- Medical Physics
- Published
- 2025-03-20
- DOI
- https://doi.org/10.1002/mp.17752
- Citations
- 1
- Primary Topic
- Interstitial Lung Diseases and Idiopathic Pulmonary Fibrosis
- Type
- article
- Field-Weighted Citation Impact
- 1.52
Funders
- National Institutes of Health