Tractomamba is a sequence-aware algorithm for whole-brain parcellation of streamlines from diffusion MRI tractography into anatomically defined white matter fiber bundles.
The core model implementation, trained model, and pretrained weights will be made publicly available after the associated paper is accepted.
Until then, these materials are withheld from the public release. The current repository contains supporting code, configuration, and label metadata only. Training and inference require the unreleased model implementation and weights and cannot be run using the public repository alone. Installation and usage instructions will be provided with the full release after paper acceptance.
configs/: model configuration.datasets/: data-loading utilities and tract label metadata.models/README.md: core model release notice.trainedmodel/README.md: trained model and weights release notice.utils/: supporting utilities.run_inference.py: inference entry point; requires the forthcoming model release.train.py: training entry point; requires the forthcoming model release.pyproject.tomlandrequirements.txt: package metadata and dependencies.
The ORG atlas used in training is available at http://dmri.slicer.org/atlases/.
The currently published files are subject to LICENSE.