Overview

Schema Version

https://atlas.rsna.org/schemas/2025-11/model.json

Name

The Scottish Medical Imaging (SMI) Archive

Link

https://web.www.healthdatagateway.org/dataset/1c49a822-6432-468b-8ba5-6aab534654b9

Indexing

Keywords: MRI, Imaging Sequences, Ultrasound, Mammography, CT, Angiography, Conventional Radiography, Scottish Medical Imaging Archive, Scottish National Safe Haven, PACS, DICOM, Trusted Research Environment

Content: IN, OT

RadLex: RID10326, RID10312, RID10321, RID10345, RID10337

Author(s)

Rob Baxter

Thomas Nind

James Sutherland

Gordon McAllister

Douglas Hardy

Ally Hume

Ruairidh MacLeod

Jacqueline Caldwell

Susan Krueger

Leandro Tramma

Ross Teviotdale

Kenny Gillen

Donald Scobbie

Ian Baillie

Andrew Brooks

Bianca Prodan

William Kerr

Dominic Sloan-Murphy

Juan F. R. Herrera

Edwin J. R. van Beek

Parminder Singh Reel

Smarti Reel

Esma Mansouri-Benssassi

Roy Mudie

Douglas Steele

Alex Doney

Emanuele Trucco

Carole Morris

Robert Wallace

Andrew Morris

Mark Parsons

Emily Jefferson

Organization(s)

EPCC, University of Edinburgh

Health Informatics Centre (HIC), University of Dundee

Public Health Scotland

Scottish National Safe Haven (hosted by University of Edinburgh for Public Health Scotland)

Health Data Research UK

NHS Scotland

Version

1.0

License

Text: CC BY 4.0

URL: https://creativecommons.org/licenses/by/4.0/

Contact

Emily Jefferson (email: ku.ca.eednud@nosreffej.r.e)

Funding

Supported by MRC (MR/M501633/1, MR/S010351/1), Wellcome Trust (WT086113), Chief Scientist Office (Scottish Government), Scottish Government “Imaging AI” grant, HDR UK (HDR-5012) with multiple funders, Edinburgh and South-East Scotland City Region Deal, University of Edinburgh, and UK charities.

Ethical review

Public Health Scotland has a generic ethics approval for a research database; PBPP approved application to bring data from national PACS into the NSH research database.

Date

Published: 2023-09-27

References

[1] Baxter R, Nind T, Sutherland J, et al.. "The Scottish Medical Imaging Archive: 57.3 Million Radiology Studies Linked to Their Medical Records". Radiology: Artificial Intelligence. 2024;6(1):e220266.. 2023-09-27. doi:10.1148/ryai.220266. PMID: 38166330. PMCID: PMC10831519.

[2] Nind T, Sutherland J, McAllister G, et al.. "An extensible big data software architecture managing a research resource of real-world clinical radiology data linked to other health data from the whole Scottish population". Gigascience. 2020;9(10):giaa095.. . PMID: 32990744. PMCID: PMC7523405. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7523405/

[3] Scottish Medical Imaging (SMI) Research Dataset. "Health Data Research UK Gateway entry". Dataset catalogue entry. 2022-11-09. Available from: https://web.www.healthdatagateway.org/dataset/1c49a822-6432-468b-8ba5-6aab534654b9

Model

Architecture

Open-source software platform and data architecture for hosting and managing population-scale radiology images linked to longitudinal health records; three-zone architecture (Secure Data Management Zone, De-identified Zone, Analysis Zone) with microservices pipeline.

Availability

Accessible within the Scottish National Safe Haven via the eDRIS SMI Service after PBPP approval: https://www.isdscotland.org/Products-and-Services/eDRIS/Scottish-Medical-Imaging-Service/

Clinical benefit

Enables health care research and development/validation of AI algorithms with population-based imaging linked to longitudinal health records in a trusted research environment.

Clinical workflow phase

Clinical research enablement within a trusted research environment; cohort assembly and data provisioning for analysis/AI development.

Degree of automation

End-to-end service for cohort extraction and data provisioning operated by eDRIS using specialized tools; not an automated diagnostic device.

Indications for use

For researchers to access pseudonymized, linkable radiology images and associated health records within the NSH to conduct health care research and develop/validate AI algorithms.

Input

Routinely collected radiology DICOM images (e.g., CT, MRI, PET, radiography, ultrasound) and linked longitudinal electronic health records (e.g., hospital, maternity, cancer registry, prescribing, mortality).

Instructions

Researchers contact eDRIS; submit PBPP application; upon approval, eDRIS analysts assemble cohorts using DICOM tags and linked clinical data; data are accessed within NSH project accounts under TRE principles; data cannot be exported.

Limitations

Scottish population has low ethnic diversity (~96% White); potential for AI models to encode personal data—disclosure control methods under research; no automated prevention yet of 3D facial reconstruction or unique identifier extraction—manual checks required; challenges stratifying images by sequence type/body part using only DICOM tags; initial archive covers 2010–2018 with ongoing updates; access constrained to NSH due to confidentiality and data protection requirements.

Output

Description: Pseudonymized, linkable cohort-specific datasets of imaging and associated health records delivered within the National Safe Haven for analysis and AI development/validation.

Recommendation

Use within the NSH with PBPP approval; subsets should match the approved research question; annotations/ground truth can be captured and shared subject to consent.

Sustainability

Managed as a national-scale resource (~3 PB of data) within a TRE; performance and resource management described in cited architecture; no specific energy metrics reported.

Use

Intended: Other

User

Intended: Other, Researcher