LongiTrack / README.md
ykirchhoff's picture
Add hf download option to README
801749e
|
Raw History Blame Contribute Delete
4.83 kB
---
license: cc-by-nc-sa-4.0
library_name: LongiSeg
pipeline_tag: image-segmentation
tags:
- longiseg
- medical-imaging
- ct
- longitudinal
- lesion-tracking
- tumor-segmentation
- segmentation
- 3d-segmentation
- autopet
- autopet-iv
- radiology
---
# LongiTrack
![LongiSeg verified lesion tracking](assets/longisegv2.gif)
LongiTrack provides point-prompted longitudinal lesion tracking and segmentation for CT. It runs through [LongiSeg](https://github.com/MIC-DKFZ/LongiSeg) using CT images and a `tracking.json` file.
Paper: [Exploiting Longitudinal Context in Clinician-Verified Interactive Lesion Tracking](https://huggingface.co/papers/2605.23118)
> **Research use only.** LongiTrack is not a medical device and must not be used as the sole basis for diagnosis, treatment, or other clinical decisions.
## Download and install
Install [uv](https://docs.astral.sh/uv/) first, then download the repository with either the Hugging Face CLI or Git LFS.
[Hugging Face CLI](https://huggingface.co/docs/huggingface_hub/guides/cli):
```bash
uv tool install huggingface_hub
hf download ykirchhoff/LongiTrack --local-dir LongiTrack
```
[Git LFS](https://git-lfs.com/):
```bash
git lfs install
git clone https://huggingface.co/ykirchhoff/LongiTrack
```
Then install:
```bash
cd LongiTrack
uv sync
```
## Run tracking inference
`tracking.json` must follow the [LongiSeg tracking-file format](https://github.com/MIC-DKFZ/LongiSeg/blob/master/documentation/lesion_tracking.md).
```bash
uv run longitrack-predict \
--images-path /path/to/images \
--output-path /path/to/output \
--tracking-path /path/to/tracking.json
```
Supported options:
| Option | Default | Description |
| --- | --- | --- |
| `--images-path` | — | Input image directory. |
| `--output-path` | — | Output directory. |
| `--tracking-path` | — | Path to `tracking.json`. |
| `--version` | `1.1` | Model version to use for inference. |
| `--mode` | `automatic` | Tracking mode: `automatic` or `manual`. |
| `--device` | `cuda` | Inference device: `cuda`, `cpu`, or `mps`. |
The selected version determines the model directory. Available checkpoint folds are detected automatically.
## Versions
| Version | Model directory | Training data |
| --- | --- | --- |
| 1.0 | `LongiTrack_v1.0/` | [AutoPET IV](https://autopet-iv.grand-challenge.org/) |
| 1.1 | `LongiTrack_v1.1/` | [AutoPET IV](https://autopet-iv.grand-challenge.org/), [PanTrack](https://huggingface.co/datasets/mrokuss/PanTrack), [HCUCH](https://zenodo.org/records/17788162) |
## Interactive viewer
[LongiTrack-napari](https://github.com/MIC-DKFZ/LongiTrack-napari) provides the interactive napari workflow for reviewing and correcting tracking points.
## Citation
### LongiTrack and LongiSeg
```bibtex
@inproceedings{kirchhoff2026exploiting,
title = {Exploiting Longitudinal Context in Clinician-Verified Interactive Lesion Tracking},
author = {Kirchhoff, Yannick and Rokuss, Maximilian and Mertens, Daniel Philipp and F{\"u}ller, David and Hamm, Benjamin and Schreyer, Andreas and Ritter, Oliver and Maier-Hein, Klaus},
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026},
year = {2026}
}
@inproceedings{rokuss2024longitudinal,
title = {Longitudinal Segmentation of MS Lesions via Temporal Difference Weighting},
author = {Rokuss, Maximilian R and Kirchhoff, Yannick and Roy, Saikat and Kovacs, Balint and Ulrich, Constantin and Wald, Tassilo and Zenk, Maximilian and Denner, Stefan and Isensee, Fabian and Vollmuth, Philipp and Kleesiek, Jens and Maier-Hein, Klaus},
booktitle = {International Conference on Medical Image Computing and Computer-Assisted Intervention},
pages = {64--74},
year = {2024},
organization = {Springer}
}
```
### Data
```bibtex
@misc{kuestner2025longitudinal,
author = {K\"ustner, Thomas and Peisen, Felix and Gatidis, Sergios and Wagner, Andreas and Megne, Ornela and Othman, Ahmed and Sanner, Antoine and Lo\ssau, Tanja and Moltz, Jan Hendrik and Kohlbrandt, Temke and Hering, Alessa},
title = {Longitudinal-CT},
year = {2025},
publisher = {University of T\"ubingen},
doi = {10.57754/FDAT.qwsry-7t837}
}
@article{rojas2026ct,
title = {A CT Dataset with RECIST Measurements and Comprehensive Segmentation Masks for Tumors and Lymph Nodes},
author = {Rojas-Pizarro, Roberto and V{\'a}squez-Venegas, Constanza and Pereira, Gonzalo and Eyssautier, Mar{\'\i}a F and Bravo-Baham{\'o}ndez, Felipe and Sanhueza, Nicol{\'a}s and Gallardo-Badilla, Paulina and Caro-Flores, Francisca and Orme{\~n}o-Candia, Camila and Santander, Felipe and others},
journal = {Scientific Data},
volume = {13},
number = {1},
pages = {270},
year = {2026},
publisher = {Nature Publishing Group UK London}
}
```