The clinical AI assistant for radiology teams. Every answer linked to its evidence.

Ask questions about imaging guidance. Review draft reports against RCR standards. Capture learning for continuing professional development (CPD).

For NHS teams, training programmes and imaging software partners.

Book a demo Try it as a clinician

Licensed NICE · RCR iRefer · BMUS · RadiopaediaUKCA Class I medical deviceNHS pilot in Newcastle

One evidence base. Three ways to use it.

Ask.

Ask anything in radiology.

Answers come from licensed UK guidance and peer-reviewed literature. Every sentence links to its source, with the strength of the evidence shown.

Example questionSuspected renal colic: which scan first?Low-dose non-contrast CT KUB is first line in adults.1 Scan within 24 hours of presentation.2 Ultrasound first in pregnancy, children and young people.3

1RCR iRefer · Suspected ureteric colic
2NICE QS195 · Renal and ureteric stones
3NICE NG118 · Renal and ureteric stones

See an answer ↓

Check.

Sharper reports, every time.

Paste an anonymised draft and get educational feedback against RCR reporting standards. Apply or dismiss each suggestion; the author stays in control.

Example suggestionSegmental vessels are not mentioned.State the extent of assessment, e.g. “No filling defect to segmental level.”ApplyDismiss

See it in action ↓

Keep.

Keep the understanding.

Log what you looked up and learnt as continuing professional development (CPD) evidence for appraisal, all in one place.

Example reflection“Next time I read a guideline, I’ll start with its scope.”Saved from: Suspected renal colic: which scan first?
CPD journal · ready for appraisal

For NHS teams, training programmes and software partners.

Clinicians use it. Organisations license it. This is what each can buy or discuss today.

NHS teamsAvailable now

Department licence

Seats for radiologists, radiographers, sonographers and trainees
Usage and answer-quality reporting for governance
Starts with a scoped pilot, measured against agreed outcomes

Discuss a pilot →

Training programmesAvailable now

Programme licence

For deaneries, universities and FRCR courses
Cited answers and CPD capture for every trainee
A view for educators of the questions being asked

Discuss a licence →

Software partnersPartnership discussions

RIS, PACS and reporting software

Available today as a secure web app and an Android app
Embedding inside a partner’s reporting workflow is scoped with each partner
We are looking for design partners

Talk integration →

Every sentence traces to a source.

Scroll, or choose a step. The answer is written in plain clinical language, then any sentence opens the guidance it came from.

MyRadAssistant

KEY POINTS

Include the clinical details, adding context where it affects interpretation of the scan.1
State the relevant findings, without irrelevant incidental findings or lists of irrelevant negatives.1
End with a short, actionable conclusion that directly answers the clinical question.1
Make clear, specific recommendations, avoiding blanket phrases such as “clinical correlation advised”.1

Tier 1 · Professional standard

Standards for interpretation and reporting of imaging investigations

The Royal College of Radiologists · Third edition, October 2025

SECTION 05 · REPORT CONCLUSION · PAGE 9“The conclusion section … is a key component of a report and should be short and actionable.” … “Directly answer the clinical question and, where relevant, show interpretation to address implied questions.”

Open original publication ↗

Sharper reports, every time.

An anonymised draft, checked against RCR reporting standards. Scroll to apply each suggestion; anything only the scan can answer stays a prompt for the author.

MyRadAssistant

ChatCase SearchReport CritiqueCPD Journal

3 to review0 for the author to confirm2 important · 1 suggested

33 RCR criteria checked · 6 guidelines cited

ORIGINAL

Report locked as source for critique

The AI only reviews what the draft says. Anything that depends on the images is left as a prompt for the author.

CTPA. The main and lobar pulmonary arteries are well opacified with no filling defect. No pulmonary embolism. Lungs clear. No pleural effusion.

ISSUES TO REVIEW

Apply suggested changes to the draft or dismiss

✓ WHAT THIS REPORT DID WELL

Clear negative impression.
Lungs and pleura reviewed.

ImportantApply all important (2)1 of 3

The findings describe only the main and lobar pulmonary arteries, but the impression states “No pulmonary embolism” without qualification.

State the most distal level assessed, then match the impression to it. Only the author can confirm the level.

ApplyDismiss

There is no statement on opacification quality or motion artefact.

Add a Technique line for the author to complete from the images.

ApplyDismiss

Suggested

Right heart strain is not commented on.

Add a prompt for the author to confirm from the images.

ApplyDismiss

WORKING DRAFT

Updated as you apply suggestions. This is the version used for export.

CT pulmonary angiogram. [Confirm opacification quality and any motion artefact.]CTPA.The main and lobar pulmonary arteries are well opacified with no filling defect.No pulmonary embolism.Lungs clear. No pleural effusion.[Comment on right heart strain after reviewing the images.]

Copy draftExportLog to CPD

Made for the whole imaging team.

Each role uses it within its own scope of practice. Choose one to see the kind of question it answers.

“Ring-enhancing brain lesion: abscess or high-grade glioma?”

A cited differential with the distinguishing features set out side by side, and the evidence tier of each source.

From the people using it.

Early users across NHS imaging teams, quoted with their permission.

“Genuinely excellent, far superior to ChatGPT or Google Gemini and even better than Navigating Radiology’s Knowledge Search… been keeping a tab with it open during every reporting session for the last week, it’s such a massive help.”

Dr Chris OwenRadiology RegistrarWest of Scotland

“During my reporting practice if I have wanted a quick answer for something I have found myself reaching for MyRadAssistant instead of [other radiology websites] or Copilot as I know it will bring up some academic sources too.”

Hefina WilliamsReporting Radiographer and NENC Imaging Academy Reporting Radiographer EducatorGateshead Health NHS Foundation Trust

“A paediatric patient was sent for a liver scan to check for haemangioma… I wasn’t sure of the guidelines, and MyRadAssistant showed the guidelines and all information I need for the scan. Saves time and also brain energy!”

Viana ShadoRadiographerNewcastle upon Tyne Hospitals NHS Foundation Trust

“Differentiating MRI appearances of arachnoid and epidermoid cysts: gave a clear concise overview with some scrollable cases. Case Search is my main everyday use, to expose myself to different appearances of the same pathology.”

Dr Mohit PatelRadiology RegistrarNewcastle upon Tyne Hospitals NHS Foundation Trust

“Genuinely excellent, far superior to ChatGPT or Google Gemini and even better than Navigating Radiology’s Knowledge Search… been keeping a tab with it open during every reporting session for the last week, it’s such a massive help.”

Dr Chris OwenRadiology RegistrarWest of Scotland

“During my reporting practice if I have wanted a quick answer for something I have found myself reaching for MyRadAssistant instead of [other radiology websites] or Copilot as I know it will bring up some academic sources too.”

Hefina WilliamsReporting Radiographer and NENC Imaging Academy Reporting Radiographer EducatorGateshead Health NHS Foundation Trust

“A paediatric patient was sent for a liver scan to check for haemangioma… I wasn’t sure of the guidelines, and MyRadAssistant showed the guidelines and all information I need for the scan. Saves time and also brain energy!”

Viana ShadoRadiographerNewcastle upon Tyne Hospitals NHS Foundation Trust

“Differentiating MRI appearances of arachnoid and epidermoid cysts: gave a clear concise overview with some scrollable cases. Case Search is my main everyday use, to expose myself to different appearances of the same pathology.”

Dr Mohit PatelRadiology RegistrarNewcastle upon Tyne Hospitals NHS Foundation Trust

Where it is used and evaluated.

Each relationship is labelled for what it is: use, evaluation, support or a completed programme.

NHS pilot at Newcastle upon Tyne HospitalsRadiology trainees use it day to day as Champion Users, feeding back on answer quality, sources and usability. Started September 2026.In use

ARROW programme, Durham UniversityDefining and validating an evaluation framework, so answer quality is measured the same way before every wider rollout.Evaluation in development

Newcastle Hospitals innovation teamSupporting the route from pilot to wider NHS adoption.Support

Mentorship from Dr Avi Mehra, IBMAssociate Partner and CSO, Healthtech. Mentoring the founding team on trustworthy AI for healthcare.Support

Evidence Generator Bootcamp, DigitalHealth.LondonProgramme on the evidence NHS buyers need, including a NICE Evidence Standards Framework gap analysis.Completed 2026

Governance and assurance.

Only what is held today, each independently verifiable.

UKCA Class I medical deviceRetrieval and answer function registered with the MHRA under UK MDR 2002.MHRA register ↗

NHS DSPT Standards MetODS U9L1B, 2025-26, valid to 30 June 2027.DSPT register ↗

ICO registeredRegistration ZB621582.ICO register ↗

Your dataQueries are never used to train models. Staff should not enter identifiable patient information.

Available on request for pilots: information governance pack, DPIA support and clinical safety case.

The people behind it.

FOUNDER

Dr Yiannis Skarparis

Consultant Interventional Radiologist

CO-FOUNDER · CLINICAL SAFETY OFFICER

Dr Ben Stenberg

Consultant Sonographer

CTO

Mr Rory Oliver

Behind GPNotebook

Questions buyers ask.

If yours is not here, ask it on the demo call.

How is this different from a general AI chatbot?

Before answering, it searches a curated radiology evidence base and answers using only that retrieved text, with every sentence linked to its source and the strength of that source shown. If the evidence is not there, it says so rather than guessing.

Does it read images?

No. It does not accept DICOM or other imaging input and does not interpret scans. It answers questions, gives feedback on the wording of draft reports and records learning.

Can it be embedded in our RIS, PACS or reporting software?

It is available today as a secure web app and an Android app. Embedding inside a partner’s workflow is scoped with each partner, and we are actively looking for design partners.

What happens to the data our staff enter?

Queries are never used to train models. Staff should not enter identifiable patient information. An information governance pack and DPIA support are available for pilots.

How would a pilot be evaluated?

We agree the teams, questions and outcomes at the start, then report back on answer quality, usage and staff feedback against them.

What happens if an answer is wrong?

Every answer has a feedback button. Safety-relevant reports go to a monitored mailbox and are triaged within five working days; serious incidents are handled under the MHRA vigilance system.

What is its regulatory status?

The retrieval and answer function is a UKCA Class I medical device under UK MDR 2002, registered with the MHRA. Report critique is educational and sits outside the device classification.

Book a demo.

Thirty minutes with a radiologist on the founding team. We show it on your kind of questions and agree whether a pilot makes sense.

Email to book

hello@myradassistant.com · we reply within two working days.

01Agree the scope.Which teams, which questions, which outcomes.
02Clear governance.IG pack, DPIA support and clinical safety case.
03Measure it.Answer quality, usage and feedback, reported back to you.

A clinician? Start a 14-day free trial, no card needed.

Coming soon