Fraxinus

An Open Source Platform for Image-Guided Bronchoscopy

Fraxinus home page — documentation, user guide, downloads, and screenshots.

Fraxinus is a pre-procedure planning and simulation platform for bronchoscopy research, developed at SINTEF Medical Technology in collaboration with St. Olavs Hospital, Trondheim University Hospital, Norway.

Lung cancer is the leading cause of cancer-related death worldwide. Pre-procedural planning, computing a route through the airway tree to the target lesion and rehearsing the procedure virtually, is a key step in improving diagnostic outcomes for peripheral pulmonary lesions.

Fraxinus provides a complete, open-source pipeline in four guided steps:

1. Patient — import a chest CT and run automatic segmentation of airways, lungs, lymph nodes, vessels, tumors, and more using AI-based tools.
2. Set Target — click in the CT to place a bronchoscopy target; the optimal route through the airway tree is computed automatically.
3. Virtual Bronchoscopy — navigate the segmented airway tree in a fly-through view synchronized with CT cross-sections, including an EBUS simulator.
4. Procedure Planning — review all segmented structures in combined 2D/3D views before entering the bronchoscopy suite.

Fraxinus is built on CustusX and uses Raidionics and TotalSegmentator for AI-based segmentation, and Elastix for image registration.

DISCLAIMER: Fraxinus is a research tool. It is not intended for routine clinical use and is neither FDA nor CE approved.

Install

Pre-built installers for Ubuntu 22.04, 24.04, and Windows are available on the releases page. Ubuntu is the primary, most thoroughly tested platform. The Windows installer uses IGSTK for tracking and has not been as extensively tested and verified as the Ubuntu builds.

On Ubuntu, install directly via curl:

sudo apt install curl -y  # optional, if curl is not already installed
curl -fsSL https://gitlab.sintef.no/custusx/fraxinus/-/releases/permalink/latest/downloads/installFraxinus.sh | bash

Alternatively, download installFraxinus.sh and run it (e.g. if you'd like to inspect the script first). If you already have an older copy, re-download it rather than running the old one -- some browsers save a repeat download as a numbered duplicate (installFraxinus (1).sh) instead of overwriting it, which would leave ./installFraxinus.sh below pointing at a stale or missing file. Each release page also lists a version-suffixed copy (e.g. installFraxinus-v26.09.sh) if you'd rather keep several versions side by side without this renaming issue:

chmod +x installFraxinus.sh
./installFraxinus.sh

The script downloads Fraxinus and installs all required inference engines and AI models (Raidionics, TotalSegmentator, Elastix) automatically.

Build from Source

Full build instructions are in the developer documentation. The basic steps are:

mkdir fx
cd fx
git clone https://gitlab.sintef.no/custusx/fraxinus.git FX/FX
cd FX/FX
./script/cxFraxinusInstaller.py --full --all --build_type Release --user_doc

Run ./script/cxFraxinusInstaller.py -h for more options.

Structure

Fraxinus is written in C++ using CMake, Qt, VTK, ITK, and other libraries, built on the CustusX platform using the CTK OSGi plugin framework. It consists of three main plugins:

• org.custusx.fraxinus — application entry point and setup logic for external AI tools
• org.custusx.fraxinus.core.state — finite state machine driving the four-step workflow and orchestrating the AI segmentation pipeline
• org.custusx.fraxinus.widgets — all user-facing interface components

Citation

If you use Fraxinus in your research, please cite:

Janne Beate Lervik Bakeng, Erlend Fagertun Hofstad, Ole Vegard Solberg, Jon Eiesland, Geir Arne Tangen, Tore Amundsen, Thomas Langø, Ingerid Reinertsen, Tormod Selbekk, Håkon Olav Leira. Using the CustusX toolkit to create an image guided bronchoscopy application: Fraxinus, PLOS ONE, 02/2019. DOI: 10.1371/journal.pone.0211772

Contributing

The source code is mirrored on GitHub. External users are welcome to submit bug reports and feature requests on GitHub Issues.

Contributors

Fraxinus is developed by SINTEF Medical Technology for St. Olavs Hospital, Trondheim University Hospital, in cooperation with the Norwegian National Research Center for Minimally Invasive and Image-Guided Diagnostics and Therapy (SINTEF, NTNU).