Evaluate AI Use in Clinical Radiology Settings
Research type
Research Study
Full title
Evaluate AI Use in Clinical Radiology Settings
IRAS ID
365938
Contact name
Hantao Liu
Contact email
Sponsor organisation
Cardiff University
Duration of Study in the UK
1 years, 1 months, 23 days
Research summary
This study aims to understand how artificial intelligence (AI) can be used safely and effectively to support radiologists when interpreting medical images such as X-rays, CT scans and MRI scans. AI tools are increasingly able to highlight abnormalities or suggest possible diagnoses. However, we still know very little about the best ways to present this information to radiologists in a real clinical setting, and how this may affect their decision-making, confidence and accuracy. This research seeks to address that gap.
The study will be carried out in collaboration with the NHS National Imaging Academy Wales. We will work closely with radiologists, human-computer interaction specialists and AI experts to design a range of different formats for presenting AI-generated information—for example, visual overlays. Radiologists will take part in user-centred design workshops to help create these presentation formats so that they are clinically useful, understandable and practical to use during busy workflows.
We will then test these prototype formats in simulated clinical environments that resemble genuine reporting conditions. Radiologists will review imaging cases with and without AI support. During this process, we will collect both quantitative data (such as reporting time, diagnostic accuracy and error rates) and qualitative data (such as user feedback and perceived usefulness).
The aim of the study is to identify which types of AI output presentation best support radiologists, improve diagnostic efficiency, and enhance patient care. The findings will lead to practical guidelines for safe and effective AI integration into radiology services. The study will also strengthen collaboration between NHS radiologists and AI researchers, helping ensure that future AI tools are designed to meet the needs of clinical practice.
REC name
N/A
REC reference
N/A