New: QCA for coronary angiography →
medrun

A model for every exam.

Upload a study, get a structured read: findings, measurements, scores and DICOM SR/SEG, organised in sections with one price per request. From the console or one API call.

One result format across 15 modalities and 110 exam types.

Example result

Structured reads, not just text.

One run returns findings, measurements, scores and an optional draft report, organised in sections. This is the stored result of a public Head CT example (Left subdural hematoma, trauma head CT).

Open this example

CQ500-CT-155.zip

Head CTCompleted

Findings

  • Intracranial hemorrhage
  • Left subdural hemorrhage
  • Midline shift 3.5 mm toward the right
  • No epidural hemorrhage

Measurements

Intracranial hemorrhage

Total hemorrhage volume
39.1 mL

Left subdural hemorrhage

Volume
39.1 mL
Maximal thickness
11.7 mm

Midline shift 3.5 mm toward the right

Septum pellucidum distance from the ideal midline
3.5 mm

Triage

critical

Intracranial hemorrhage

Report

Left subdural hemorrhage measuring 11.7 mm in maximal thickness with 3.5 mm midline shift toward the right. Total intracranial hemorrhage volume is 39.1 mL. No ventriculomegaly.

One library,
from the cath lab
to the slide scanner

Explore all models
  1. Upload

    Drop a DICOM folder, a CD export or a zip. Folders keep their structure; DICOMDIR is honoured.

  2. Read

    MedRun detects the exam, runs the model and streams progress, critical flags first. The result comes in sections.

  3. Integrate

    A queue API with signed webhooks and an MCP server. Results come back as JSON, DICOM SR/SEG, PDF and FHIR.

Three ways
to use MedRun.

Console

Drop a CD, get a read.

Drop a DICOM folder, a CD export or a zip. MedRun suggests the model and streams the result as it runs.

  • Cine and image viewers with overlays
  • Request history, usage and credits
  • Teams, roles and API keys
Open the console

API & integrations

One API for every exam.

A queue API with server-sent progress, signed webhooks and an MCP server.

  • Submit, follow and fetch in three calls
  • Webhooks for completed and critical results
  • MCP tools for Claude and other assistants
Read the docs

MedRun models

Exam-level reads.

Every model returns one complete result in sections: findings, measurements, scores, guidance and an optional draft report.

  • A second read for every study
  • Quantified, guideline-structured findings
  • Masks, overlays and key frames
Explore models

Why
MedRun?

Segment%DSMLD mmmLAD62.00.77pLCx38.51.20

trace → measurements/m-L1-pct

Every number traces to a finding

Values carry their unit, method, uncertainty and the frame they came from. The draft report is validated against them.

TID 1500SEGCAD-RADSTI-RADSLung-RADSBI-RADSFHIR R4IHE AIR

Standards are the floor

Each section follows its reporting standard. Results go back to PACS and EHRs as DICOM SR/SEG and FHIR.

const req = await medrun.run("angio/coronary", {
  files: ["file_01K6…"],
  options: { report: true },
});
for (const s of req.output.sections)
  console.log(s.id, s.status);

Built for developers

One queue API, server-sent progress, signed webhooks and generated OpenAPI per endpoint.

Documentation

One price per request

No per-section pricing. Prepaid credits with auto top-up; rejected inputs are not charged.

See pricing

Built for hospitals

Teams, roles and an audit log; pinned model versions and provenance on every result; short retention of working copies.

  • DICOM in, DICOM out
  • Teams & roles
  • Audit log
  • Signed webhooks
  • Pinned versions
  • 7-day retention

Read your next study with MedRun

Start in the console with your own files, or call the API from your product.