Tutorials¶

The demos of LION, as tutorials: each builds on the last, from the CT operator to publishing a trained model, and together they cover most of LION’s API. Each is a Python file with # %% cells, to run whole, a cell at a time in an editor that knows cells (VS Code, Spyder), or as a notebook (download it below each page).

They need a GPU (LION’s CT operators use tomosipo and ASTRA) and, from the data tutorial on, the processed LIDC-IDRI data in LION_DATA_PATH.

The models trained in the tutorials are toys, trained for a few epochs on a few slices to show how LION is used: their results are poor. Trained as their papers train them, they do what the papers report.