TamRx
Menu

TamRx Ltd · Leeds, United Kingdom

Clinical AI should show its evidence, not just its answer.

TamRx is developing decision support for difficult breast ultrasound cases. Instead of a single score, it presents similar cases with confirmed outcomes, and defers to the clinician when the evidence is insufficient.

Join a reader study How it works

Two similar-looking breast ultrasound lesions, one benign and one malignant, both confirmed by biopsy.
Two biopsy-confirmed breast lesions that look alike on ultrasound: one benign, one malignant. Images: BrEaST-Lesions_USG (The Cancer Imaging Archive), CC BY 4.0.
Development status: TamRx software is a research prototype in development. It is not UKCA marked or FDA cleared, and is not available for clinical use.

The problem

A flag is not an explanation.

In breast screening and assessment, AI flags and look-alike lesions create uncertainty. Some lesions look benign and are not; others look suspicious and are not. Clinicians need evidence they can inspect, not a black-box score they are asked to trust.

Our vision

Fewer missed and mistaken diagnoses, with the clinician always in charge.

For clinicians

A second look grounded in confirmed cases, so a difficult call can be checked against the evidence rather than a number.

For patients

Fewer unnecessary recalls and biopsies, and fewer cancers that look benign going unrecognised.

For the system

AI that can be inspected, audited and trusted because it shows its reasoning and states its limits.

How it works

Similar cases. Confirmed outcomes. A visible differential.

The approach follows the way specialists reason: consider the plausible diagnoses, weigh the evidence, and recognise when a case needs more than the available evidence can give.

  1. Similar confirmed cases

    For a selected region, TamRx retrieves the most similar reference cases, with outcomes confirmed by histology or follow-up.

  2. Evidence for and against

    Similar cases are grouped by diagnosis, showing what supports and what argues against each candidate, so the clinician can see what remains plausible.

  3. Reliability signal

    When the closest confirmed cases disagree, or no sufficiently similar case exists, TamRx reports insufficient evidence and defers to the clinician.

Research & evidence

Measured carefully, reported honestly

  • Leakage-controlled benchmark. We are building a benchmark across public breast ultrasound datasets, with patient-level splits and duplicate control, so that results are not overstated.
  • Research collaborations. Research collaborations with UK and European academic groups are in discussion.
  • Compute allocation. TamRx has been awarded 10,000 GPU hours on the Leonardo Booster supercomputer (CINECA, Italy) through the EuroHPC AI Factory Fast Lane programme, for our research project on computer vision for ambiguous and rare breast lesions.
  • Clear about our current status. We will publish methods and results once they have been independently reviewed.

For clinicians

Help shape responsible imaging AI

We are inviting breast radiologists, radiographers, sonographers and breast clinicians to test the approach on difficult cases and tell us where it helps and where it falls short. Your judgement sets the standard it has to meet.

  • One to two hours, online
  • Public data only
  • No patient data is shared or requested

Reserve a reader-study place

I am interested in

Stored securely in the UK. See our privacy notice.

Clinical safety & governance

Governance from the start

  • Data protection. Registered with the Information Commissioner's Office (ZC266873). We process personal data in line with UK GDPR.
  • Data governance. Development uses de-identified, licence-checked public datasets. No identifiable patient data is processed.
  • Clinical safety. Clinical risk management is being prepared in line with DCB0129. A Clinical Safety Officer will be appointed before any NHS deployment.
  • Quality and regulation. We are working towards ISO 13485-aligned quality management and the UKCA and FDA pathways.
  • Status. Research prototype. Not UKCA marked or FDA cleared. Not for clinical use.
  • Intellectual property. UK patent application filed.

Governance & compliance · Clinical safety (DCB0129) · Security

Team

Portrait of Mounsif Tamani

Mounsif Tamani

Founder & CEO

Trained in pharmacy (PharmD) and data science (MSc Biological Data Science). Innovate UK Expert Assessor in medical AI and health technology. Mounsif founded TamRx to give clinicians transparent, evidence-based support for the cases that are hardest to call.

Jobs

Build responsible imaging AI with us

We are forming a small founding team and are recruiting two senior roles.

Head of NHS Commercialisation

Part-time initially · Founding-team role · UK

Lead TamRx's commercial route into the NHS, from first evaluations to procurement and adoption. Experience of NHS procurement, DTAC and DCB0129/DCB0160, with an understanding of MHRA device regulation and ISO 13485.

ApplyFull description

Clinical Lead, Breast Imaging

Part-time · a few hours per month · Remote / Leeds

Be the clinical voice in how TamRx is designed, evaluated and kept safe. A consultant radiologist or senior clinician with breast imaging and ultrasound experience.

ApplyFull description

Both roles are currently remunerated with vesting equity, moving to paid roles after our seed round.

Programmes & support

Programmes and support

  • MemberNexus, University of Leeds: Member
  • MemberNVIDIA Inception: Member
  • ComputeEuroHPC AI FactoryLeonardo supercomputer allocation (CINECA)
  • Supported byAWS ActivateAmazon Web Services
  • ProgrammesHealth Innovation Yorkshire & HumberPropel and IEP
  • MemberNashville Entrepreneur CenterMember

Contact

We welcome conversations with clinicians, academic groups, educators and investors interested in evidence-based decision support for breast imaging.

mounsif@tamrx.co.uk +44 7426 668241

Nexus, Discovery Way, Leeds LS2 3AA, United Kingdom