CIL on the Cloud

CIL-Demos

CIL-Demos is a collection of jupyter notebooks, designed to introduce you to the Core Imaging Library (CIL).

The demos can be found in the demos folder, and the README.md in this folder provides some info about the notebooks, including the additional datasets which are required to run them.

Binder

To open and run the notebooks interactively in an executable environment, please click the Binder link above.

Note: In the Binder interface, there is no GPU available.

Note: In the Google Cloud platform, there is free GPU (16Gb). However, you need to install CIL manually.

Install the demos locally

To install via conda, create a new environment using:

conda create --name cil -c conda-forge -c https://software.repos.intel.com/python/conda -c ccpi cil=24.2.0 ipp=2021.12 astra-toolbox=*=cuda* tigre ccpi-regulariser tomophantom ipykernel ipywidgets scikit-image

where,

astra-toolbox will allow you to use CIL with the ASTRA toolbox projectors (GPLv3 license).

tigre will allow you to use CIL with the TIGRE toolbox projectors (BSD license).

ccpi-regulariser will give you access to the CCPi Regularisation Toolkit.

tomophantom Tomophantom will allow you to generate phantoms to use as test data.

cudatoolkit If you have GPU drivers compatible with more recent CUDA versions you can modify this package selector (installing tigre via conda requires 9.2).

ipywidgets will allow you to use interactive widgets in our jupyter notebooks.

Dependency Notes

CIL's optimised FDK/FBP recon module requires:

1. the Intel Integrated Performance Primitives Library (license) which can be installed via conda from the https://software.repos.intel.com/python/conda channel.
2. TIGRE, which can be installed via conda from the ccpi channel.

Run the demos locally

• Activate your environment using: conda activate cil-demos.

• Clone the CIL-Demos repository and move into the CIL-Demos folder.

• Run: jupyter-notebook on the command line.

• Navigate into demos/1_Introduction

The best place to start is the 01_intro_walnut_conebeam.ipynb notebook. However, this requires installing the walnut dataset.

To test your notebook installation, instead run 03_preprocessing.ipynb, which uses a dataset shipped with CIL, which will have automatically been installed by conda.

Instead of using the jupyter-notebook command, an alternative is to run the notebooks in VSCode.

Advanced Demos

For more advanced general imaging and tomography demos, please visit the following repositories:

• Core Imaging Library part I: a versatile python framework for tomographic imaging

• Core Imaging Library part II: multichannel reconstruction for dynamic and spectral tomography.