🐭 LuTuFlow

LuTuFlow is a Python toolbox for segmenting and tracking lung tumor nodules in longitudinal series of mice CT scans. It is developed as a collaboration between the EPFL Center for Imaging and the De Palma Lab.

Hightlights

• Detect the lungs automatically using a pretrained YoloV8 model and crop the scans around them.
• Detect tumor nodules using a pretrained nnUNet 3D segmentation model.
• Track individual tumors across several CT scans of the same mouse.

Installation

As a standalone app

Download and run the latest installer from the Releases page. This is the simplest option, but it only allows usage in Napari (not as a CLI).

In Python

We recommend performing the installation in a clean Python environment. First, install the zeroc-ice package via the pre-built wheels from Glencoe Software. Choose the wheel corresponding to your python version (3.10, 3.11, 3.12) and platform (Windows, MacOS, Linux).

• Windows: https://github.com/glencoesoftware/zeroc-ice-py-win-x86_64/releases/
• MacOS: https://github.com/glencoesoftware/zeroc-ice-py-macos-universal2/releases/
• Linux: https://github.com/glencoesoftware/zeroc-ice-py-linux-x86_64/releases

Then, install our package from PyPi:

pip install lutuflow

or from the repository:

pip install git+https://github.com/EPFL-Center-for-Imaging/lutuflow.git

or clone the repository and install with:

git clone git+https://github.com/EPFL-Center-for-Imaging/lutuflow.git
cd lutuflow
pip install -e .

Usage

LuTuFlow can be used in Napari or from the command-line. See the documentation for usage instructions.

License

This project is licensed under the AGPL-3 license.

This project depends on the ultralytics package which is licensed under AGPL-3.

This project uses the PyApp software for creating a runtime installer.

Citing

If you use LuTuFlow in the context of scientific publication, please cite our preprint:

BibTeX

@article {Liu2026.10.12.758509,
	author = {Liu, Yuchen and Li, Luqing and Wittwer, Mallory and Martinez-Usatorre, Amaia and Mansouri, Nahal and Dapia, Bastien and Ando, Edward and De Palma, Michele},
	title = {LuTuFlow: a deep learning-based pipeline for quantitative characterization of mouse lung tumors in CT imaging},
	elocation-id = {2026.10.12.758509},
	year = {2026},
	doi = {10.64898/2026.10.12.758509},
	URL = {https://www.biorxiv.org/content/early/2026/10/14/2026.10.12.758509},
	journal = {bioRxiv}
}