What the survey found
Researchers set out to understand how patients actually feel about AI in their imaging care, and published the results in RSNA's journal Radiology. The team surveyed more than 1,000 patients in the waiting areas of an imaging department between July 2024 and April 2025.
The headline number is hard to ignore: about 96% said patients should be informed when AI is used to report on their images. That's not a slim majority — it's near-unanimous. AI disclosure in imaging is no longer a hypothetical policy question; it's an expectation patients already hold.
They also want it to be explicit. Asked how they'd prefer to consent, 53% chose written consent versus 34% who would accept verbal consent. Patients aren't satisfied with AI use being buried in a form they never see — a majority want it documented.
The accountability signal everyone should notice
The most revealing finding wasn't about disclosure at all. When asked who would be at fault if an incorrect result was produced when images were read by AI alongside a radiologist, 64% said both the doctor and the technology.
Read that closely. Patients don't imagine a future where AI reads the scan alone and takes the blame alone. They assume a human radiologist stays in the loop and shares responsibility for the final result. Their mental model of trustworthy AI in imaging already includes a physician who is accountable — which is exactly the model regulators and professional bodies endorse, since no AI system today is approved for autonomous final reporting without radiologist sign-off.
Why this is now an operational question, not an ethics seminar
For years, "AI transparency in healthcare" lived in conference panels. This survey drags it into daily operations. If 96% of your patients would want to know that AI touched their report, then how you disclose it — and whether you can credibly say a radiologist reviewed every read — becomes part of your patient experience, not a footnote.
Three practical implications follow for imaging centers, teleradiology providers, and hospital imaging departments:
Disclosure becomes part of intake
With a majority preferring written consent, AI use is drifting toward the same explicit-consent posture as other parts of care. Centers adopting AI should decide now how they surface it at intake, rather than retrofitting it after a patient asks.
A radiologist must stay accountable — visibly
Patients assign shared fault to the physician and the tool. That only holds if a radiologist actually reviews and signs the report. AI that produces a draft a radiologist reviews fits this expectation; fully autonomous reads do not.
Trust is a competitive variable
As AI use becomes routine, being able to say clearly "AI assists, a radiologist signs, and we tell you" is a differentiator with patients and referrers — not just a compliance checkbox.
Where this fits with how AI CT reporting actually works
The survey describes the model patients already trust: AI assists, a radiologist stays accountable. That's the model AI CT reporting is built on — the AI produces a structured, comprehensive report draft, and a radiologist reviews and signs every one before it reaches a patient's chart. Every xAID report is radiologist-reviewed by design. The transparency patients are asking for isn't a constraint on this workflow — it's a description of it.
Frequently asked questions
Do patients want to know when AI is used to read their scan?
Yes. In a 2026 survey of more than 1,000 imaging patients published in RSNA's journal Radiology, about 96% said patients should be informed when AI is used to report on their images. The survey was conducted in the waiting areas of an imaging department between July 2024 and April 2025 in Ireland.
How do patients want to consent to AI use in imaging?
In the same survey, 53% of respondents preferred to give consent in writing, while 34% preferred verbal consent. This suggests patients want disclosure to be explicit and documented, not implicit — a practical consideration for how imaging centers integrate AI into intake and reporting workflows.
Who do patients think is responsible if an AI-assisted report is wrong?
When asked who would be at fault if an incorrect result was provided when images were read by AI together with a radiologist, 64% of respondents said both the doctor and the technology. Patients intuitively expect a human radiologist to remain accountable for the final report, even when AI assists.
What does patient disclosure mean for imaging centers using AI?
It means AI adoption is not only a clinical and financial decision but a trust decision. Centers should be prepared to disclose when AI assists a report, keep a radiologist accountable for every final read, and communicate that human oversight clearly. Doing so aligns with what patients say they expect and reduces the risk of eroding trust as AI use becomes routine.
Source: survey of imaging patients published in RSNA's Radiology (2026), as reported by Radiology Business and AuntMinnie. Figures are rounded as reported.