Scottish Medical Imaging (SMI) – providing safe and secure access to research-ready, population scale health and imaging data

  • Cole, C. (Contributor)
  • Krueger, S. (Speaker)
  • Jackie Caldwell (Contributor)
  • Robert Wallace (Contributor)
  • Ruairidh MacLeod (Contributor)

Activity: Talk or presentation typesInvited talk

Description

Routinely collected text-based health care records and associated imaging data are extremely useful for healthcare research. However, using such data is challenging and requires specialist tools and skills to work with. Trusted Research Environments (TREs) such as Scottish Medical Imaging (SMI) provide the capabilities and controls needed to unlock the value of such data whilst ensuring patient confidentiality.

We share our solutions to some of the major research challenges faced in establishing SMI and making imaging data research-ready, through the PICTURES collaboration:

Cataloguing – the metadata that describes each image, series or study is often inconsistent, incomplete or incorrect, making it very difficult to reliably index and catalogue image data for cohort building.
Natural Language Processing – useful image descriptors can be found in associated Radiology Reports, specifically free-text notes describing expert observations and diagnosis.
Pixel anonymisation – many scanning devices or secondary processing steps result in Personally Identifiable Information (PII) being burnt-in to the images themselves, requiring tools and methods for identifying and removing PII from pixel data at-scale.
Period5 Sept 2023
Event titleResearch Software Engineering Conference 2023
Event typeConference
Conference number7
LocationSwansea, United KingdomShow on map