Triage Radiology Imaging Assessment for Greater Effectiveness

Official title

Triage Radiology Imaging Assessment for Greater Effectiveness (TRIAGE): a Cluster-Randomised Crossover Trial Evaluating the Reporting Turn-Around-Time of Brain Magnetic Resonance Images Using an AI-enabled Abnormality Detector for Prioritisation

Overview

The aim of the trial is to understand whether a computerised tool designed to quickly spot problems in brain scans (MIDI) can help doctors diagnose cases faster. This will help patients by getting them the treatment they need sooner. How useful doctors find the tool will also be measured, and whether it is cost-saving for the National Health Service (NHS).

Detailed description

NHS radiology services are under unprecedented pressure as rising demand outpaces available workforce capacity. MRI utilisation across the NHS has increased by 9% year-on-year. As a result, many departments have relied on unsustainable measures such as outsourcing and weekend insourcing to manage reporting backlogs.

Delays in scan reporting have serious implications, including missed or delayed diagnoses, prolonged hospital stays, higher healthcare costs, and poorer patient outcomes. A systematic review and meta-analysis found that each month of delay in initiating cancer treatment increases mortality risk by approximately 10%.

There is an urgent need for validated digital tools that can reduce TATs without compromising diagnostic accuracy. The TRIAGE study (Triage Radiology Imaging Assessment for Greater Effectiveness (TRIAGE): a Cluster-Randomised Crossover Trial Evaluating the Reporting Turn-Around-Time of Brain Magnetic Resonance Images using an AI-enabled Abnormality Detector for Prioritisation) evaluates whether MIDI can reduce the average TAT of abnormal scans in routine NHS clinical practice.

The primary purpose of MIDI is workflow triage: it enables prioritisation of abnormal scans within radiology reporting worklists, helping departments manage high volumes efficiently.

Interventions

• Device MIDI
  MIDI is an AI-based TRIAGE tool and its primary purpose is workflow triage of brain MRI scans.

Primary outcome measures

• Turnaround time for abnormal outpatient brain MRI scans [Time frame: Time from abnormal MRI acquisition to radiologist report authorisation over 8 months.]

Secondary outcome measures (12)

• Turnaround time for abnormal inpatient brain MRI scans [Time frame: Time from abnormal MRI acquisition to radiologist report authorisation over 8 months.]
• Turnaround time for abnormal combined Brain MRI scans [Time frame: Time from abnormal MRI acquisition to radiologist report authorisation over 8 months.]
• Cost Effectiveness - Incremental cost per reduction in brain MRI scan turnaround time, radiologist reporting time, insourcing/outsourcing of reporting activity [Time frame: From MRI scan acquisition to final radiologist report authorisation over 8 months.]
• Cost Effectiveness for patients with suspected TIA or Stroke - incremental cost per reduction in brain MRI turnaround time, time to diagnosis and time to treatment change [Time frame: From MRI scan acquisition to initiation of treatment over 8 months.]
• Cost Effectiveness for patients newly diagnosed with brain metastases secondary to melanoma, lung cancer, or breast cancer [Time frame: From MRI scan acquisition to initiation of treatment within 8 months as well as from MRI scan acquisition to 28 and 90 days for mortality and 90 days for ECOG/KPS outcomes.]
• Staff satisfaction - Questionnaire score from radiologists, radiographers, and referring clinicians, collected at month 6 of the intervention [Time frame: From the start to month 6 of the intervention.]
• Turnaround time for all brain MRI scans (normal and abnormal combined) [Time frame: Time from MRI acquisition to final radiologist report acquisition over 8 months]
• Turnaround Time for Critically Abnormal Brain MRI scans [Time frame: Acquisition of critically abnormal brain MRI scans to final radiologist report authorisation over 8 months.]
• Performance accuracy - MIDI assessment of abnormal/normal brain MRI scan, radiologist assessment of abnormal/normal brain MRI scan and reference standard of abnormal/normal brain MRI scan [Time frame: MRI scan acquisition to final radiologist report authorisation over 8 months.]
• Clinical Effectiveness Outcomes in patients with Suspected Transient Ischemic Attack/Stroke (time from MRI acquisition to definitive diagnosis, censoring or treatment change) [Time frame: Time from MRI Acquisition to Definitive Diagnosis or Censoring, Time from MRI Acquisition to Treatment Change during Phase I and II (19 months)]
• Proportion of diagnoses within 14 days in a subgroup of patients diagnosed with melanoma or lung cancer with de novo metastases [Time frame: Within 14 days following index brain scan containing brain metastases]
• Mortality rates in a subgroup of patients diagnosed with melanoma or lung cancer with de novo metastases [Time frame: Within 28 days following the index MRI brain scan and within 90 days following the index MRI brain scan]

Eligibility criteria

Inclusion criteria

For main trial

• Adult patient (≥ 18 years of age) at the time of MRI.
• Brain MRI scan performed at a participating NHS site during the defined study period (intervention or control arm).
• Brain MRI scan series will automatically be included if they contain the required scan sequences and are of the required resolution.

For O3 subgroup ● Radiologists, radiographers, and referring clinicians involved with the use of the MIDI tool at sites who consent to completing the staff survey

For O7 subgroup

• Adult patient (≥ 18 years of age) at the time of MRI.
• Brain MRI scan performed at a participating NHS site during the defined study period (intervention or control arm).
• Brain MRI scan series will automatically be included if they contain the required scan sequences and are of the required resolution.
• Patient with suspected de novo stroke/TIA in TIA/Stroke pathway.
• Patient with suspected de novo stroke/TIA in Acute Neurology pathway.
• Patient with suspected de novo stroke/TIA identified via SSNAP review.

For O8 subgroup

• Adult patient (≥ 18 years of age) at the time of MRI.
• Brain MRI scan performed at a participating NHS site during the defined study period (intervention or control arm).
• Brain MRI scan series will automatically be included if they contain the required scan sequences and are of the required resolution.
• Patient with stage III/IV malignant melanoma undergoing regular surveillance MRI brain scans presenting with de novo brain metastases.
• Patient with lung cancer undergoing pre-radical treatment staging MRI brain scans presenting with de novo brain metastases.
• Patients with de novo symptomatic brain metastases secondary to melanoma and/or lung cancer and/or breast cancer detected on MRI brain scans from all patient pathways.

Additionally for those completing EORTC questionnaires:

● Able to give consent and likely to be able to complete EORTC questionnaires.

Exclusion criteria

For main trial

• Patient is under 18 years of age at the time of MRI.
• Brain MRI scan not performed at a participating NHS site or outside the defined study period.
• Data will be excluded for those patients who have opted out of the use of their data for research under the National Data Opt-Out.
• Brain MRI scan series will automatically be excluded if they do not contain the required scan sequences and/or the required resolution.

For O3 subgroup ● Radiologists, radiographers, and referring clinicians involved with the use of the MIDI tool at sites who do not consent to completing the staff survey

For O7 subgroup

• Patient is under 18 years of age at the time of MRI.
• Brain MRI scan not performed at a participating NHS site or outside the defined study period.
• Data will be excluded for those patients who have opted out of the use of their data for research under the National Data Opt-Out.
• Brain MRI scan series will automatically be excluded if they do not contain the required scan sequences and/or the required resolution.
• Patient without suspected de novo stroke/TIA or not on TIA/Stroke pathway.
• Patient without suspected de novo stroke/TIA or not on Acute Neurology pathway.

For O8 subgroup

• Patient is under 18 years of age at the time of MRI.
• Brain MRI scan not performed at a participating NHS site or outside the defined study period.
• Data will be excluded for those patients who have opted out of the use of their data for research under the National Data Opt-Out.
• Brain MRI scan series will automatically be excluded if they do not contain the required scan sequences and/or the required resolution.
• Patient without stage III/IV malignant melanoma, or not undergoing regular surveillance MRI brain scans, or not presenting with de novo brain metastases.
• Patient without lung cancer undergoing pre-radical treatment staging MRI brain scans, or not presenting with de novo brain metastases.
• Patients without de novo symptomatic brain metastases secondary to melanoma and/or lung cancer and/or breast cancer detected on MRI brain scans.

Additionally for those completing EORTC questionnaires:

● Unable to give consent and unlikely to be able to complete EORTC questionnaires.

Criteria are shown verbatim from the registry (in English). Final eligibility is always assessed by the study center.

Healthy volunteers: No

Study design

Allocation

Randomized

Model

Crossover

Masking

Open label

Primary purpose

Diagnostic

Study locations

Center list to be confirmed — check the primary protocol.

Publications

• Candel MJ, van Breukelen GJ. Sample size calculation for treatment effects in randomized trials with fixed cluster sizes and heterogeneous intraclass correlations and variances. Stat Methods Med Res. 2015 Oct;24(5):557-73. doi: 10.1177/0962280214563100. Epub 2014 Dec 17. PMID 25519890
• van Breukelen GJ, Candel MJ. Calculating sample sizes for cluster randomized trials: we can keep it simple and efficient! J Clin Epidemiol. 2012 Nov;65(11):1212-8. doi: 10.1016/j.jclinepi.2012.06.002. PMID 23017638
• Hooper R, Teerenstra S, de Hoop E, Eldridge S. Sample size calculation for stepped wedge and other longitudinal cluster randomised trials. Stat Med. 2016 Nov 20;35(26):4718-4728. doi: 10.1002/sim.7028. Epub 2016 Jun 28. PMID 27350420
• Powers DMW (2011). "Evaluation: From Precision, Recall and F-Measure to ROC, Informedness, Markedness & Correlation". Journal of Machine Learning Technologies. 2 (1): 37-63. doi:10.1186/s12880-015-0068-x.
• ClinicalTrials.gov. Study NCT04368481. https://clinicaltrials.gov/study/NCT04368481https://www.sciencedirect.com/science/article/pii/S1361841522000433
• Wood DA, Kafiabadi S, Busaidi AA, Guilhem E, Montvila A, Lynch J, Townend M, Agarwal S, Mazumder A, Barker GJ, Ourselin S, Cole JH, Booth TC. Deep learning models for triaging hospital head MRI examinations. Med Image Anal. 2022 May;78:102391. doi: 10.1016/j.media.2022.102391. Epub 2022 Feb 12. PMID 35183876
• NHS England (2023). Faster Diagnosis Standard. https://www.england.nhs.uk/cancer/faster-diagnosis
• NICE (2022). Evidence Standards Framework for Digital Health Technologies. https://www.nice.org.uk/corporate/ecd7

Identifiers

NCT: NCT07671664 · 350683

Primary sources (government registries)

• ClinicalTrials.gov (US NIH)

View this study on ClinicalTrials.gov ↗