Towards generalist foundation model for radiology by leveraging web-scale 2D&3D medical data

Abstract In this study, as a proof-of-concept, we aim to initiate the development of Radiology Foundation Model, termed as RadFM. We consider three perspectives: dataset construction, model design, and thorough evaluation, concluded as follows: (i), we contribute 4 multimodal datasets with 13M 2D im...

Full description

Saved in:
Bibliographic Details
Main Authors: Chaoyi Wu, Xiaoman Zhang, Ya Zhang, Hui Hui, Yanfeng Wang, Weidi Xie
Format: Article
Language:English
Published: Nature Portfolio 2025-08-01
Series:Nature Communications
Online Access:https://doi.org/10.1038/s41467-025-62385-7
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849226120009875456
author Chaoyi Wu
Xiaoman Zhang
Ya Zhang
Hui Hui
Yanfeng Wang
Weidi Xie
author_facet Chaoyi Wu
Xiaoman Zhang
Ya Zhang
Hui Hui
Yanfeng Wang
Weidi Xie
author_sort Chaoyi Wu
collection DOAJ
description Abstract In this study, as a proof-of-concept, we aim to initiate the development of Radiology Foundation Model, termed as RadFM. We consider three perspectives: dataset construction, model design, and thorough evaluation, concluded as follows: (i), we contribute 4 multimodal datasets with 13M 2D images and 615K 3D scans. When combined with a vast collection of existing datasets, this forms our training dataset, termed as Medical Multi-modal Dataset, MedMD. (ii), we propose an architecture that enables to integrate text input with 2D or 3D medical scans, and generates responses for diverse radiologic tasks, including diagnosis, visual question answering, report generation, and rationale diagnosis; (iii), beyond evaluation on 9 existing datasets, we propose a new benchmark, RadBench, comprising three tasks aiming to assess foundation models comprehensively. We conduct both automatic and human evaluations on RadBench. RadFM outperforms former accessible multi-modal foundation models, including GPT-4V. Additionally, we adapt RadFM for diverse public benchmarks, surpassing various existing SOTAs.
format Article
id doaj-art-257cc54748da4cf88f97428bd98bd46b
institution Kabale University
issn 2041-1723
language English
publishDate 2025-08-01
publisher Nature Portfolio
record_format Article
series Nature Communications
spelling doaj-art-257cc54748da4cf88f97428bd98bd46b2025-08-24T11:38:14ZengNature PortfolioNature Communications2041-17232025-08-0116112210.1038/s41467-025-62385-7Towards generalist foundation model for radiology by leveraging web-scale 2D&3D medical dataChaoyi Wu0Xiaoman Zhang1Ya Zhang2Hui Hui3Yanfeng Wang4Weidi Xie5Shanghai Jiao Tong UniversityShanghai Jiao Tong UniversityShanghai Jiao Tong UniversityShanghai Jiao Tong UniversityShanghai Jiao Tong UniversityShanghai Jiao Tong UniversityAbstract In this study, as a proof-of-concept, we aim to initiate the development of Radiology Foundation Model, termed as RadFM. We consider three perspectives: dataset construction, model design, and thorough evaluation, concluded as follows: (i), we contribute 4 multimodal datasets with 13M 2D images and 615K 3D scans. When combined with a vast collection of existing datasets, this forms our training dataset, termed as Medical Multi-modal Dataset, MedMD. (ii), we propose an architecture that enables to integrate text input with 2D or 3D medical scans, and generates responses for diverse radiologic tasks, including diagnosis, visual question answering, report generation, and rationale diagnosis; (iii), beyond evaluation on 9 existing datasets, we propose a new benchmark, RadBench, comprising three tasks aiming to assess foundation models comprehensively. We conduct both automatic and human evaluations on RadBench. RadFM outperforms former accessible multi-modal foundation models, including GPT-4V. Additionally, we adapt RadFM for diverse public benchmarks, surpassing various existing SOTAs.https://doi.org/10.1038/s41467-025-62385-7
spellingShingle Chaoyi Wu
Xiaoman Zhang
Ya Zhang
Hui Hui
Yanfeng Wang
Weidi Xie
Towards generalist foundation model for radiology by leveraging web-scale 2D&3D medical data
Nature Communications
title Towards generalist foundation model for radiology by leveraging web-scale 2D&3D medical data
title_full Towards generalist foundation model for radiology by leveraging web-scale 2D&3D medical data
title_fullStr Towards generalist foundation model for radiology by leveraging web-scale 2D&3D medical data
title_full_unstemmed Towards generalist foundation model for radiology by leveraging web-scale 2D&3D medical data
title_short Towards generalist foundation model for radiology by leveraging web-scale 2D&3D medical data
title_sort towards generalist foundation model for radiology by leveraging web scale 2d 3d medical data
url https://doi.org/10.1038/s41467-025-62385-7
work_keys_str_mv AT chaoyiwu towardsgeneralistfoundationmodelforradiologybyleveragingwebscale2d3dmedicaldata
AT xiaomanzhang towardsgeneralistfoundationmodelforradiologybyleveragingwebscale2d3dmedicaldata
AT yazhang towardsgeneralistfoundationmodelforradiologybyleveragingwebscale2d3dmedicaldata
AT huihui towardsgeneralistfoundationmodelforradiologybyleveragingwebscale2d3dmedicaldata
AT yanfengwang towardsgeneralistfoundationmodelforradiologybyleveragingwebscale2d3dmedicaldata
AT weidixie towardsgeneralistfoundationmodelforradiologybyleveragingwebscale2d3dmedicaldata