Pipeline guide · v1 draft

An open-source imaging AI loop: Orthanc → mercure → TotalSegmentator → OHIF

This is a conceptual reference stack made of four widely used open-source projects. It describes what each component is, based on the projects’ own descriptions. Follow each project’s official docs for installation and configuration. This is for research and development environments only, not for clinical use.

1. Store: Orthanc

A lightweight, open-source DICOM server that acts as your research PACS. The orthanc-builder repo builds the official Docker images, Windows installers and macOS packages. License: AGPL-3.0.

orthanc-server.com · orthanc-builder

2. Route: mercure

A DICOM orchestrator: a routing and processing solution with a web interface and monitoring, used to send incoming studies to processing modules such as AI models. License: MIT.

github.com/mercure-imaging/mercure

3. Process: TotalSegmentator

Segments more than 100 anatomical structures in CT and MR images. License: Apache-2.0 (check the repo for model and subtask terms).

github.com/wasserth/TotalSegmentator

4. View: OHIF Viewer

A zero-footprint, web-based DICOM viewer for reviewing studies and results. License: MIT.

github.com/OHIF/Viewers · docs.ohif.org

Step-by-step deployment instructions (Docker Compose, DICOMweb wiring, sending results back as DICOM SEG) will be added after we test them hands-on. We don’t publish steps we haven’t verified.