Unimodal imaging foundation models and vision-language models in MRI: A systematic review of emerging methods and clinical applications.
AI models show promise reading brain MRIs, though automated report writing still needs work; standardized datasets and safety checks will enable real clinical translation.
This systematic review of 35 studies evaluated unimodal foundation models and multimodal vision-language models applied to MRI across neuroimaging, oncology, and musculoskeletal domains, finding neuroimaging diagnostic classification to be the most mature application. Automated MRI report generation remains clinically limited despite linguistic fluency, and advancing clinical translation requires standardized multicenter datasets, electronic health record integration, and robust safety evaluation frameworks.
What the study was
- Study design
- systematic_review
- Category
- ai_ml_clinical_diagnostics
- Maturity
- Validated
- Journal
- Cell Reports Medicine
Why it surfaced
First comprehensive systematic review of FMs and VLMs in MRI; published in Cell Reports Medicine; identifies gaps in clinical translation and priorities for future development.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.