Can AI write reports like a radiologist? A blinded evaluation of large language model-generated lumbar spine MRI reports
LLM-generated reports are clinically coherent and stylistically comparable to those written by expert radiologists. Radiologist-written reports scored significantly higher for clinical relevance, findings, and structure. LLM-generated reports were sometimes misclassified as human-written by clinicians.
Authors
- Renato Fabrizio (ORCID: https://orcid.org/0009-0000-9066-8749)
- Moreno Zanardo (ORCID: https://orcid.org/0000-0001-9640-8534)
- Francesco Traina (ORCID: https://orcid.org/0000-0002-0196-101X)
- Luigi Asmundo (ORCID: https://orcid.org/0000-0002-1410-341X)
- Martina Conca
- Valentina Molinari
- Salvatore Gitto
- Domenico Albano
- Francesco Pardo
- Michele Montechiari
Institutions
- University of Milan (IT)
- Azienda Socio Sanitaria Territoriale Grande Ospedale Metropolitano Niguarda (IT)
- IRCCS Policlinico San Donato (IT)
- IRCCS Ospedale Galeazzi - Sant'Ambrogio (IT)
- Istituto Ortopedico Rizzoli (IT)
- ASST Fatebenefratelli Sacco (IT)
- University of Bologna (IT)
Publication Details
- Journal
- European Radiology Experimental
- Published
- 2026-02-23
- DOI
- https://doi.org/10.1186/s41747-026-00682-6
- Primary Topic
- Radiology practices and education
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Ministero della Salute