Intracranial Hemorrhage Detection and Subtype Classification on CT Imaging Using a Large Language Model
The LLM demonstrated limited sensitivity for hemorrhage detection and inconsistent subtype classification, underscoring current constraints of zero-shot application to medical imaging. These findings highlight the need for domain-specific fine-tuning, larger and more diverse datasets, and integration with traditional computer vision methods before clinical deployment.
Authors
- Rahul Kumar (ORCID: https://orcid.org/0000-0002-8372-9627)
- Alireza Tavakkoli (ORCID: https://orcid.org/0000-0001-9460-1269)
- T. M. Praneeth Naidu (ORCID: https://orcid.org/0000-0002-7029-6308)
- Andrew Bouras (ORCID: https://orcid.org/0000-0002-7960-9358)
- Ethan Waisberg (ORCID: https://orcid.org/0000-0001-8999-0212)
- Nitin Chetla (ORCID: https://orcid.org/0009-0003-3496-262X)
- Shivam Patel
Institutions
- Philadelphia College of Osteopathic Medicine (US)
- University of Nevada, Reno (US)
- University of Massachusetts Chan Medical School (US)
- University of Cambridge (GB)
- Cambridge School (PT)
- University of Virginia (US)
- Virginia Tech (US)
Publication Details
- Journal
- Cureus
- Published
- 2025-11-28
- DOI
- https://doi.org/10.7759/cureus.98029
- Primary Topic
- Intracerebral and Subarachnoid Hemorrhage Research
- Type
- article
- Field-Weighted Citation Impact
- 0.00