Triaging Radiographs for Reporting Using Artificial Intelligence
Research type
Research Study
Full title
TRAIN: Triaging Radiographs for Reporting Using Artificial Intelligence
IRAS ID
291727
Contact name
Giovanni Montana
Contact email
Sponsor organisation
University of Warwick
Duration of Study in the UK
0 years, 3 months, days
Research summary
Clinical need: Imaging underpins most medical decisions to the point that radiology utilisation has outstripped resourcing in the UK. Current radiology workforce shortages are contributing to delays in reporting. There is a clear need to address if there are better ways of triaging imaging for prompt reporting.
Hypothesis: We propose that a computer-based ‘artificial intelligence’ (AI) system could be used to better select chest radiographs (X-rays) for urgent reporting than current methods in real-time so that the time to reporting for urgent findings (irrespective of clinical history or referrer) can be reduced.
Our prior research: Our team have developed an AI system to detect chest X-ray abnormalities (Annarumma et al. Radiology 2019; 291(1):196-202) and have shown in a simulation its potential to reduce delays to reporting critical findings.
Current proposal aims: We wish to test if our AI system can triage chest X-rays as they are taken in real time; by detecting important findings, our AI system can highlight X-rays that need reporting more urgently, improving reporting turnaround times for radiologists.
The following will be assessed:
1. Changes to prospective reporting times when this triaging system is followed by on-site radiologists.
2. Triaging performance for critical, urgent, non-urgent and normal chest X-rays versus expert reviewImpact on patient care:
We hope that our system can reduce reporting time for chest X-rays. In this research project, no other change in the usual patient care is anticipated.REC name
South East Scotland REC 02
REC reference
21/SS/0028
Date of REC Opinion
23 Apr 2021
REC opinion
Further Information Favourable Opinion