trustmed AI

Medical image annotation, powered by autonomous AI agents.

Intelligent model selection, automated segmentation and independent quality verification across medical imaging domains.

Try it free Contact us

Start with $10 of free credit. No card needed.

Multi-modal

Built for X-ray, CT, MRI, ultrasound, fundus, dermoscopy, histology and angiography, with a specialist agent for each imaging domain.

Model-agnostic

The right model for every task: specialised segmentation models and foundation models, chosen by the agents, with a library that keeps growing.

Human-in-the-loop

An independent reviewer checks every mask. Uncertain or low-confidence results are flagged for expert review instead of slipping through.

How it works

From an upload to a verified annotation, in six steps

A team of AI agents does the work an annotation team would: read the image, pick the tools, draw, check, and send anything uncertain to a person.

  1. Image upload

    Upload your images and say what to annotate, in plain words.

  2. Image understanding

    An agent identifies the modality, the anatomy and the annotation task.

  3. Model selection

    The orchestration layer picks the most suitable segmentation model for the task.

  4. Automated annotation

    Models draw the masks; agents refine them piece by piece.

  5. Independent verification

    A separate reviewer checks each mask, names its errors and sends it back to be fixed.

  6. Human review and export

    Uncertain cases go to human experts; accepted results are ready to export.

A real run

From medical images to verified annotations

A public chest X-ray and one request: Outline the pneumothorax. The reviewer rejected the first two masks and said why; the third was accepted. Nothing here is retouched.

A frontal chest X-ray
The image as uploaded
The first mask in pink, reaching into the lung below the pleural line
First mask Rejected · 0.68
The final mask on its own, white on black
Final mask as delivered
The final mask in green over the X-ray, following the pleural line
Verified Accepted · 0.90

What the agents read

Request
Outline the pneumothorax
Category
Chest X-ray
Modality
X-ray
View
Frontal, likely PA
Image quality
Good
Guided by
5 expert-outlined examples from other patients

Independent review

  1. Round 1 · rejected 0.68 The lung-side boundary extends below the visible visceral pleural line into vascularized left upper lung.
  2. Round 2 · rejected 0.78 Structure 1 includes a substantial band of vascularized upper-left lung.
  3. Round 3 · accepted 0.90 The mask covers the left apical and lateral pneumothorax, broadly following the visceral pleural boundary and inner chest wall.

Image and expert examples: PTX-498 (Wang et al.), CC BY 4.0; masks drawn by trustmed AI. Scores are the reviewer's own, from 0 to 1.

Who it's for

For teams that need annotated medical images

Medical AI startups

Label training data faster, with every mask checked before it reaches your model.

Biomedical research labs

Turn image collections into annotated datasets without months of manual outlining.

Universities and researchers

Annotate across modalities for studies and theses, with the reasoning behind every mask.

Imaging AI developers

Start a new task from a handful of expert examples and grow specialised models from it.

Medical device and healthtech companies

Every step recorded and reviewable, with experts in the loop where it matters.

Privacy and security

Privacy by design

  • Encrypted. In transit and at rest.
  • Private to you. Other users never see your images or results.
  • Built for de-identified data. Please upload de-identified or public images only.
  • No tracking. Only the cookies that keep you signed in.

Contact

Questions? Talk to us

A pilot on your data, a demo, a research collaboration, or anything else: the team answers by email. Or write to [email protected].

Please don't send patient data or images here.