At this year’s UKIO, one panel in particular stood out to our team — a lively and thought-provoking discussion on autonomous AI in radiology. Representing Oxipit, my colleague Vytautas Naujalis and I felt especially seen. After all, Oxipit still offers the only CE-certified solution for autonomous reporting of High Confidence Normal chest X-rays. We’ve been in this space for years, and it’s encouraging to see the conversation starting to catch up.

The debate made one thing clear: autonomous reporting in the UK isn’t currently permitted under IR(ME)R — unless certain conditions are met. If a radiologist is able to sign off on an auto-generated report and there is a robust quality assurance process in place, then the concerns seem to fade. This stands in stark contrast to current practice, where – according to the RCR 2024 Workforce Survey – 33% of UK radiology departments are either auto-reporting or not reporting certain studies at all due to pressure on capacity.

We believe there’s a better way. And more importantly — it already exists.

The Oxipit CXR Suite doesn’t just remove normal cases from the radiologist’s worklist. It does so with 99.9% specificity. It supports diagnostic decision-making for the remaining studies with on-screen assistance, and enables built-in analytics for quality assurance. It doesn’t bypass radiologists — it works with them. Our autonomous reports can be routed to a non-priority queue for review, signed if desired, and include disclaimers tailored to the institution’s patient communication preferences and regulatory needs.

This isn’t theoretical. This is in use.

So, here’s the real question: Either we have a workforce crisis, or we don’t. And if we do — and the RCR’s data makes it clear we do — then technology must be part of the solution.

Removing normal chest X-rays from the workload is a logical first step. Zero clicks. No extra interface. Just an algorithm running quietly in the background, ruling out over 75 pathologies, generating the report, and updating the worklist in the RIS or PACS. Seamless, efficient, and safe.

At Oxipit, we’ve started to view debates like the one at UKIO not as a question of whether autonomous technology should be used, but how. The technology is autonomous — but the workflow is in your hands. Want to sign the report? Add a disclaimer? Review a QA dashboard? You can. Just like our customers outside the UK already do.

Perhaps the IR(ME)R barrier is not a blocker — but a misunderstanding, or a mirage — slowing the adoption of a much-needed tool.

We were also glad to see other innovations tackling the workload challenge featured at UKIO. Solutions like xWave Technologies, embedding iRefer guidelines directly into referrer workflow, and SmartReporting, where structured reports generated with AI and speech recognition helped increase reporting volume by 32% without increasing workload. Smart tools, real impact.

Between these conversations and the hard truths laid out in the RCR 2024 Workforce Survey, one thing becomes increasingly clear: the technology is ready. What remains is the will — and the practical steps — to put it into use.

If that means allowing a signature on an auto-generated report, it seems a small price to pay for a more sustainable, patient-safe radiology service.

Peter Corscadden
CEO, Oxipit