Explainable AI in nuclear medicine
We conclude that explanations are likely required when a black box AI tool undertakes tasks that users cannot perform or validate themselves. If the user can verify outputs manually, documented reliability and accuracy may suffice, but explainability can still add value when outputs are uncertain or errors occur. More generally, close collaboration among physicians, AI developers, and other stakeholders is crucial to ensure that AI tools are trustworthy and useful in clinical practice.
Authors
- Shaul A. Duke (ORCID: https://orcid.org/0000-0003-2670-480X)
- Osama Sabri (ORCID: https://orcid.org/0000-0002-6425-3504)
- Josef Yu (ORCID: https://orcid.org/0000-0003-1426-1538)
- Sune Holm (ORCID: https://orcid.org/0000-0002-3812-7942)
- Stefan Grünert
- Lalith Kumar Shiyam Sundar (ORCID: https://orcid.org/0000-0003-3004-7937)
- Peter Sandøe (ORCID: https://orcid.org/0000-0003-0397-3273)
- Elisabetta Abenavoli (ORCID: https://orcid.org/0000-0002-9370-102X)
- Thomas Bøker Lund (ORCID: https://orcid.org/0000-0001-5282-1562)
- Bernhard Hennig (ORCID: https://orcid.org/0000-0002-7200-302X)
- Marcus Hacker (ORCID: https://orcid.org/0000-0002-4222-4083)
- Roberto Sciagrà (ORCID: https://orcid.org/0000-0003-0694-2552)
- Armin Frille (ORCID: https://orcid.org/0000-0003-2843-3863)
- Daria Ferrara (ORCID: https://orcid.org/0000-0002-8711-8081)
- Thomas Beyer (ORCID: https://orcid.org/0000-0002-8308-3918)
- Swen Hesse (ORCID: https://orcid.org/0000-0002-2055-2764)
- Miriam Pepponi
- Lukas Hofmann (ORCID: https://orcid.org/0000-0002-8397-1819)
Institutions
- University of Copenhagen (DK)
- Azienda Ospedaliero-Universitaria Careggi (IT)
- University Hospital Leipzig (DE)
- Medical University of Vienna (AT)
- Leipzig University (DE)
Publication Details
- Journal
- European Journal of Nuclear Medicine and Molecular Imaging
- Published
- 2025-11-25
- DOI
- https://doi.org/10.1007/s00259-025-07675-4
- Primary Topic
- Artificial Intelligence in Healthcare and Education
- Type
- article
- Field-Weighted Citation Impact
- 0.00