Med Frontier

Med Frontier Suppliers & Consultancy UK

Specialist radiology for clinical trials and medical AI

Med Frontier provides specialist radiology reading for imaging CROs and reference-standard datasets for medical AI companies. Reads are blinded, agreement is measured, disagreements are adjudicated by rules fixed in advance. You receive locked, traceable outputs — not opinions. We also build Radiologx, a doctor-in-the-loop reporting assistant for radiology teams.

Axial CT scan of the thorax in lung window, showing both lungs, the airways and the vertebral body
Axial CT of the thorax, lung window. Image by Ptrump16 via Wikimedia Commons, CC BY-SA 4.0 — cropped; this crop is offered under the same licence. Full image credits.
2 business lines: clinical trial imaging services and radiology AI
≥2 independent blinded reads behind every reader-derived reference standard we build
7 pre-specified steps from claim definition to locked dataset
14562381 Companies House company number, England & Wales — est. 2022, Leeds

Two lines of work

What we do

Product — in pilot evaluation

Radiologx Workflow Copilot

A doctor-in-the-loop AI assistant for oncology and MSK radiology, built for NHS radiology workflows. Structured report drafting, follow-up comparison and interval-change alerts — inside the workflow radiologists already use.

  • AI prepares; the doctor decides and signs
  • Every output reviewed, owned and auditable
  • Currently in pilot evaluation
Radiologx →

Services — for CROs & imaging AI teams

Imaging Reading & Reference-Standard Services

Independent blinded reading, pre-specified adjudication and documented reference standards — delivered by licensed specialist radiologists with senior hospital appointments, working under your charter and your quality system.

  • Subcontracted specialist reads for imaging CROs
  • Reference standards for AI validation sets
  • Agreement statistics and the full disagreement record
Imaging Services →

How we work

TruthRule™: a method you can audit, step by step

TruthRule™ is our seven-step reference-standard method. We define the claim type with you, then approve and version-lock the reading protocol. Two or more radiologists read independently and blinded; we measure their agreement and adjudicate differences by rules fixed in advance. The dataset is then locked and handed over with its full provenance. The protocol, adjudication charter and a worked example on public data are all available to review.

Review our method and worked example →