AI-powered portable MRI abnormality detection

  • Research type

    Research Study

  • Full title

    AI-powered portable MRI abnormality detection

  • IRAS ID

    347453

  • Contact name

    Thomas C Booth

  • Contact email

    thomas.booth@kcl.ac.uk

  • Sponsor organisation

    King’s College London

  • Duration of Study in the UK

    3 years, 0 months, 1 days

  • Research summary

    Our objective is to develop and validate an artificial intelligence (AI) tool for triaging portable magnetic resonance imaging (MRI) brain scans in adult patients into "normal" and "abnormal".

    Recent technology allows MR imaging using a small, portable MRI scanner that does not require a dedicated hospital room, is safe to use next to metalwork, and is much cheaper to buy and run than standard fixed MRI. While the resulting images are not as clear as scans from fixed MRI scanners, they are of sufficient clarity for relatively simple tasks such as sorting patients into normal and abnormal. If abnormal, patients can be prioritised for onward referral for standard (fixed) MRI where the superior images can be used for more complex assessments.

    Additionally, we previously developed an AI tool for triage of standard MRI brain scans, which was trained using >100,000 MRI scans acquired on a fixes scanner. This tool allows radiologists to report abnormal brain scans preferentially before normal scans. In turn, this results in faster management of patients, reduced effects of disease, and lower healthcare costs.

    In this study we will use an AI approach called “transfer learning” to combine knowledge from the existing AI tool for triage with scans from the portable scanner to build a triage tool for portable MRI.

    We will first train the model on 250 scans, test its performance on 63 new scans, and finally validate its performance using 63 scans from a separate portable MRI scanner.

    The research has immense potential to contribute to hospital and community medicine. The portable MRI scanner can plausibly be used in GP surgeries, Community Diagnostics Centres and at the bedside of patients in an Intensive Care Unit. Where an AI triage tool would allow more rapid onward referral for targeted imaging in those with an abnormality.

  • REC name

    South West - Frenchay Research Ethics Committee

  • REC reference

    24/SW/0154

  • Date of REC Opinion

    27 Jan 2025

  • REC opinion

    Further Information Favourable Opinion