Patient perceptions of artificial intelligence in MRI reconstruction: A mixed methods study from a single NHS trust.
Authors
Affiliations (7)
Affiliations (7)
- Barts Health NHS Trust, United Kingdom. Electronic address: [email protected].
- Barts Health NHS Trust, United Kingdom. Electronic address: [email protected].
- Barts Health NHS Trust, United Kingdom. Electronic address: [email protected].
- Barts Health NHS Trust, United Kingdom. Electronic address: [email protected].
- Barts Health NHS Trust, United Kingdom. Electronic address: [email protected].
- Guys and St Thomas' NHS Trust, United Kingdom; King's College London, United Kingdom. Electronic address: [email protected].
- Barts Health NHS Trust, United Kingdom; University of Manchester, United Kingdom. Electronic address: [email protected].
Abstract
As part of a strategy to increase magnetic resonance imaging (MRI) capacity, the UK government funded the purchase of manufacturer artificial intelligence (AI) MRI image reconstruction software on MRI scanners. It aimed to reduce scan duration without compromising image quality and has been widely installed on clinical MRI scanners across the UK. This is a mixed methods study, using an initial questionnaire to quantify opinions followed by focus groups of responders to further explore their views on technical issues. Patient views on the use of AI in MRI image reconstruction for reducing acquisition times, and more generally in their MRI pathway, were assessed. Study participants prefer improved image quality over faster acquisition times. If reducing acquisition times with AI, they feel it should only be used if extremely accurate images can be ascertained. They are uncomfortable with the use of manufacturer AI software for image reconstruction where accuracy is not stated and would like to be told about the use of AI in MRI. Many participants felt that manufacturer AI software for MRI image reconstruction may be used only if a high accuracy is demonstrated, and if the training data is representative of the patient population, to not exacerbate health inequalities. This information is not currently given by manufacturers, which made the patient representatives concerned.