LION Documentation¶
LION is a PyTorch framework for learned tomographic and computational-imaging reconstruction. It combines CT geometry and operators, reusable datasets and experiments, learned reconstruction models, classical baselines, and reproducible research pipelines.
This first documentation release gives full attention to the PaDIS diffusion workflows, the LIDC-IDRI extensions used by them, and the associated CT reconstruction tools. Older package areas are indexed so that their public surface is discoverable, but are explicitly marked where narrative documentation is still incomplete.
Install LION, configure data paths, and verify the environment.
Understand the package layers and how data, models, solvers, and reconstructors interact.
Train priors, tune reconstruction, run the experiment matrix, and produce paper artefacts.
Browse the documented Python interfaces and provisional package stubs.
Quick start¶
Install LION in an isolated Conda environment and point it at a data root:
conda env create --file env_base.yml --name lion
conda activate lion
pip install -e ".[dev]"
export LION_DATA_PATH=/path/to/Data
Build this site locally with:
cd docs
make html
User guide
Reference