Threshold optimization in AI chest radiography analysis: integrating real-world data and clinical subgroups
Customizing AI thresholds individually addresses specific user/patient subgroup needs. The presented approach utilizes pathology-enriched and real-world subgroup data for optimization. Potential is shown by comparing individualized thresholds with vendor defaults. Distinct thresholds for in- and outpatient CXR AI analysis may improve perception. The automated pipeline methodology is transferable to other AI models or subgroups.
Authors
- Nicola Fink (ORCID: https://orcid.org/0000-0002-3089-9606)
- Thomas Liebig (ORCID: https://orcid.org/0000-0002-9841-1101)
- Sophia S. Goller (ORCID: https://orcid.org/0000-0002-0964-8386)
- Christian Huemmer (ORCID: https://orcid.org/0000-0002-4341-4217)
- Boj Hoppe (ORCID: https://orcid.org/0000-0001-6248-5128)
- Jens Ricke (ORCID: https://orcid.org/0000-0001-8762-8594)
- Vincent Schwarze (ORCID: https://orcid.org/0000-0003-3910-7801)
- Bastian O. Sabel (ORCID: https://orcid.org/0000-0001-6027-7309)
- Jan Rudolph (ORCID: https://orcid.org/0000-0002-4849-8034)
- Giulia Buizza (ORCID: https://orcid.org/0000-0001-5504-1262)
- Johannes Rueckel
- Maurice M. Heimer (ORCID: https://orcid.org/0000-0002-0940-0161)
- Julien Dinkel (ORCID: https://orcid.org/0000-0003-1239-6084)
- Alexander Preuhs (ORCID: https://orcid.org/0000-0001-9588-2488)
- Vanessa Koliogiannis
Institutions
- Siemens (Germany) (DE)
- LMU Klinikum (DE)
- Asklepios Fachkliniken München-Gauting (DE)
- German Center for Lung Research (DE)
- Siemens Healthineers (Germany) (DE)
- Ludwig-Maximilians-Universität München (DE)
Publication Details
- Journal
- European Radiology Experimental
- Published
- 2025-09-22
- DOI
- https://doi.org/10.1186/s41747-025-00632-8
- Primary Topic
- COVID-19 diagnosis using AI
- Type
- article
- Field-Weighted Citation Impact
- 0.00