Self-Supervised Text-Vision Alignment for Automated Brain MRI Abnormality Detection: A Multicenter Study (ALIGN Study)

Purpose To develop a self-supervised text-vision framework to detect abnormalities on brain MRI scans by leveraging free-text neuroradiology reports, eliminating the need for expert-labeled training datasets. Materials and Methods This retrospective and prospective multicenter study included 81 936 brain MRI examinations and corresponding radiology reports for adult patients at two UK National Health Service hospitals from January 2008 to December 2019 for training and internal testing and 1369 prospectively collected examinations between March 2022 and March 2024 from four separate National Health Service hospitals for external testing (ClinicalTrials.gov no. NCT04368481). A neuroradiology language model (NeuroBERT) was trained using self-supervised tasks to generate report embeddings. Convolutional neural networks (one per MRI sequence) were trained to map scans to embeddings by minimizing mean squared error loss. The framework then detected abnormalities in new examinations by scoring scans against query sentences using text-image similarity. Model diagnostic performance was assessed using the area under the receiver operating characteristic curve (AUC). Results The framework achieved an AUC of 0.95 (95% CI: 0.94, 0.97) for normal versus abnormal classification and generalized to external sites with examination-level AUCs of 0.90 (95% CI: 0.86, 0.93) in Bedford, 0.87 (95% CI: 0.83, 0.90) in Nottingham, 0.86 (95% CI: 0.83, 0.90) in Norwich, and 0.85 (95% CI: 0.81, 0.89) in Yeovil. In five zero-shot classification tasks-acute stroke, multiple sclerosis, intracranial hemorrhage, meningioma, and hydrocephalus-the framework achieved a mean AUC of 0.89 (range, 0.77-0.93). For visual-semantic image retrieval, mean precision was 0.84 among the top 15 images across seven pathologies. Conclusion The self-supervised text-vision framework accurately detected brain MRI abnormalities without expert-labeled datasets. Clinical trial registration no. NCT04368481 <b>Keywords:</b> Head and Neck, Unsupervised Learning, Convolutional Neural Network (CNN), Neuroradiology © The Author(s) 2025. Published by the Radiological Society of North America under a CC BY 4.0 license. <i>Supplemental material is available for this article.</i> See also commentary by Ghodasara in this issue.

Authors

Institutions

Publication Details

Journal
Radiology Artificial Intelligence
Published
2025-11-26
DOI
https://doi.org/10.1148/ryai.240619
Citations
1
Primary Topic
Intracerebral and Subarachnoid Hemorrhage Research
Type
article
Field-Weighted Citation Impact
1.39
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Self-Supervised Text-Vision Alignment for Automated Brain MRI Abnormality Detection: A Multicenter Study (ALIGN Study)

James H. Cole, Thomas C. Booth, R. K. Nichols, Nina Mansoor et al.
1 citations
Radiology Artificial Intelligence
Intracerebral and Subarachnoid Hemorrhage Research
1.39
article

Self-Supervised Text-Vision Alignment for Automated Brain MRI Abnormality Detection: A Multicenter Study (ALIGN Study)

James H. Cole, Thomas C. Booth, R. K. Nichols, Nina Mansoor, Kishan Dissanayake, Sobha Xavier P, Carmen Dragos, David Wood, Ayisha Al Busaidi, Peter David Sasieni, Kavi Fatania, Robert A. Dineen, Maria Filyridou, Carolyn Costigan, Permesh Singh Dhillon, Emily Guilhem, Gehad Abdalla, Hilmar Spohr, Janak Saada, A. Geetha, Arne Juette, Pirrone-Brusse Michael, Martin Lewis, Kanika Bhatia, Sina Kafiabadi, Tharunniya Vamadevan, Ahmed Hammam, Stuart Currie, Giuseppe Manfredi, Gareth J. Barker, Matthew Townend, Sébastien Ourselin, Siddharth Agarwal, Mark Igra, Miguel Bertoni, Nikhil Nair, Yiran Wei, Asif Mazumder, Helen Estall, Muthu Magesh, Maria Pantelidou, Jeremy Macmullen-Price, Chike Onyekwuluje, Berna Aygun
article en
1 citations

Abstract

Purpose To develop a self-supervised text-vision framework to detect abnormalities on brain MRI scans by leveraging free-text neuroradiology reports, eliminating the need for expert-labeled training datasets. Materials and Methods This retrospective and prospective multicenter study included 81 936 brain MRI examinations and corresponding radiology reports for adult patients at two UK National Health Service hospitals from January 2008 to December 2019 for training and internal testing and 1369 prospectively collected examinations between March 2022 and March 2024 from four separate National Health Service hospitals for external testing (ClinicalTrials.gov no. NCT04368481). A neuroradiology language model (NeuroBERT) was trained using self-supervised tasks to generate report embeddings. Convolutional neural networks (one per MRI sequence) were trained to map scans to embeddings by minimizing mean squared error loss. The framework then detected abnormalities in new examinations by scoring scans against query sentences using text-image similarity. Model diagnostic performance was assessed using the area under the receiver operating characteristic curve (AUC). Results The framework achieved an AUC of 0.95 (95% CI: 0.94, 0.97) for normal versus abnormal classification and generalized to external sites with examination-level AUCs of 0.90 (95% CI: 0.86, 0.93) in Bedford, 0.87 (95% CI: 0.83, 0.90) in Nottingham, 0.86 (95% CI: 0.83, 0.90) in Norwich, and 0.85 (95% CI: 0.81, 0.89) in Yeovil. In five zero-shot classification tasks-acute stroke, multiple sclerosis, intracranial hemorrhage, meningioma, and hydrocephalus-the framework achieved a mean AUC of 0.89 (range, 0.77-0.93). For visual-semantic image retrieval, mean precision was 0.84 among the top 15 images across seven pathologies. Conclusion The self-supervised text-vision framework accurately detected brain MRI abnormalities without expert-labeled datasets. Clinical trial registration no. NCT04368481 <b>Keywords:</b> Head and Neck, Unsupervised Learning, Convolutional Neural Network (CNN), Neuroradiology © The Author(s) 2025. Published by the Radiological Society of North America under a CC BY 4.0 license. <i>Supplemental material is available for this article.</i> See also commentary by Ghodasara in this issue.

Radiology Artificial IntelligenceVol. 8(2)
Leeds General Infirmary (GB), Nottingham University Hospitals NHS Trust (GB), University of Nottingham (GB), Wellcome Centre for Human Neuroimaging (GB), Queen Mary University of London (GB), King's College London (GB), Guy's and St Thomas' NHS Foundation Trust (GB), Norfolk and Norwich University Hospital (GB), Yeovil District Hospital NHS Foundation Trust (GB), King's College Hospital NHS Foundation Trust (GB), Genomics England (GB), Bedford Hospital (GB), University College London (GB)
Openalex Percentile: Top 14%
Intracerebral and Subarachnoid Hemorrhage Research
1.39
DOIOpen Access PDFOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.