NICE backs five AI prostate MRI tools for funded NHS use while evidence is gathered

NICE has issued draft guidance that would let the NHS use five AI tools for reading prostate MRI scans, with NHS funding, during a two-year evidence-generation period ¹. The tools are AI-Rad Companion Prostate MR, DeepHealth Prostate, ProstatID, Prostate Intelligence and QP-Prostate ¹. Two further products, hProstate and mdProstate, were not backed, and NICE said they need more research ¹.

What NICE decided

All seven products do the same job: they mark regions of a prostate MRI scan where cancer may be present, so that radiologists can find and report suspicious areas more easily ¹. NICE's support is conditional. Over the next two years, each of the five supported tools must show how accurately it performs in routine NHS practice ¹. The vendors must also show its effect on the number of biopsies, the wait for biopsy and diagnosis, radiologist capacity and workflow, and patient outcomes ¹.

Dr Anastasia Chalkidou, NICE's programme director for healthtech, said the tools could serve as “a second pair of eyes, helping radiologists review prostate MRI scans” ¹. She added that they could make reporting more accurate and “could help identify people whose scans need priority review” ¹.

The clinical context explains the interest. Prostate cancer accounts for around 58,000 diagnoses a year, according to Prof Prabhakar Rajan of Barts Cancer Institute, Queen Mary University of London, and MRI sits at the centre of its diagnosis and management ¹. NICE notes that many prostate tumours can be monitored rather than treated, so MRI helps clinicians decide who needs a biopsy or further tests ¹.

What it means for the market

For NHS imaging departments and cancer alliances, the draft guidance would give a funded route to adopt any of the five products now, rather than waiting for a full assessment ¹. That narrows the field. Vendors outside the list, including the two that NICE said need more research, start at a disadvantage in NHS procurement for prostate MRI support ¹.

The evidence-generation model also shifts work onto suppliers and the trusts that host them. Each supported product now has two years to produce real-world data on accuracy and on service measures such as biopsy volumes, waiting times and radiologist workload ¹. Those results will decide whether the tools win a lasting recommendation, so the contracts that trusts sign in this period will double as evaluation sites.

Caveats

This is draft guidance, and conditional use depends on new evidence rather than on proven benefit ¹. Rajan said it is important to separate good performance at flagging or excluding suspicious findings from “evidence of improved patient outcomes” ¹. He said the open questions are whether the tools reliably find clinically significant cancers, cut unnecessary biopsies and overdiagnosis of indolent tumours, and work fairly across patient groups and NHS settings ¹.

The draft guidance and its evidence generation plan are now open for comment ¹.