ConTEXTual Net 3D: Vision-Language Modeling in PET/CT for Visual Grounding of Positive Findings
F-fluciclovine (F1 = 0.66) exams. In conclusion, our novel weak labeling pipeline accurately produced an annotated dataset of PET/CT image-text pairs. ConTEXTual Net 3D significantly outperformed other models but fell short of the performance of nuclear medicine physicians. Our study suggests that even larger datasets may be needed to close this performance gap.
Authors
- Xin Tie (ORCID: https://orcid.org/0000-0003-3062-5995)
- Zachary Huemann
- Tyler Bradshaw (ORCID: https://orcid.org/0000-0001-9549-7002)
- Steve Y. Cho (ORCID: https://orcid.org/0000-0003-2724-7778)
- Joshua Warner
- Meghan G. Lubner (ORCID: https://orcid.org/0000-0003-3788-2871)
- Junjie Hu (ORCID: https://orcid.org/0000-0002-8483-5501)
- Samuel Church (ORCID: https://orcid.org/0000-0002-5499-523X)
- Daniel Tran
- Alan B. McMillan
Institutions
- University of Wisconsin–Madison (US)
- University of Wisconsin Carbone Cancer Center (US)
Publication Details
- Journal
- Journal of Imaging Informatics in Medicine
- Published
- 2026-04-14
- DOI
- https://doi.org/10.1007/s10278-026-01879-2
- Primary Topic
- Multimodal Machine Learning Applications
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- National Institute of Biomedical Imaging and Bioengineering