CIL "Bring Your Own Data" Training and Hackathon, Cambridge 20-23 March 2023

Venue

• Isaac Newton Institute (INI) and Centre for Mathematical Sciences (CMS)
• Rooms to be confirmed

CMS - https://goo.gl/maps/BL4w2ZLQLfwcvE127

Accommodation

• Rooms reserved at Clare College

• CCPi sponsored participants should have received confirmation of room allocation

• Booking code for other attendees should be emailed - you will have to book and pay for your own room using the supplied booking code

• A 15 minute walk between venues https://goo.gl/maps/aYbFv627cFMbYiUP6

• Check in at Clare College Memorial Court:

  • Check in at the Porter's lodge after 3:00pm. The Porter's Lodge is manned 24/7 so late check-in is fine
  • There is a luggage room on site which guests are welcome to use
  • Check out is 10:00am

Goals

• allow new users to use CIL for their tomographic imaging data processing,
• help users develop readers for their data
• help users try different reconstruction algorithms for their data
• build an open collection of interesting and different use cases of CIL, for different types of data, different reconstruction methods, etc.

Open science collection of data and demos

In addition to helping individual participants bring home working CIL scripts for their data, we hope - by combining all the outcomes - to build an open collection of interesting and different use cases of CIL, for different types of data, different reconstruction methods, etc. In this spirit we invite participants to

1. bring data sets that have or can be released to the public, for example on zenodo.org,
2. develop a nice demo (script of jupyter notebook) that demonstrates the CIL-based processing of their data set, new idea or which ever topic has been worked on in the hackathon (each demo will list its contributors)

This collection will be an excellent community-built resource showcasing the depth and breadth of things that can be done with CIL and serve as inspiration and guide for new users of CIL and tomography/inverse problems researchers in general.

It will also be an extreme useful outcome for us to point to when we are to apply for funding for the continued development of CIL in the near future, so please help us help you! :-)

User contributions to CIL

We are very keen to help participants become contributors to CIL! If you have found a bug, can make the documentation clearer, or have a fantastic new feature or tool you want to add to CIL - great! We are open to contributions and can help you make your first pull request with proposed new code into the CIL codebase. Both use of CIL and contributions to CIL are covered by the Apache 2.0 License: https://choosealicense.com/licenses/apache-2.0/

Timetable

Discord server

We have set up a dedicated channel for the hackathon on our CIL Discord server to allow participants to chat and collaborate easily in a Slack style chat. Join the CIL Discord server here:
discord.gg/9NTWu9MEGq

Cloud setup

We are working on setting up a cloud platform with GPU access as an option for participants to use during the hackathon, please stay tuned!

Possible subthemes

We believe the hackathon will be most efficient - and enjoyable for participants - if working together in small groups towards goals shared by group members.

We are organizing the below list of ideas for subthemes to help participants identify shared interests and help form small groups and set goals.

Participants: Feel free to

• add your name under the subtheme(s) that you are interested in,
• add links and brief description of data sets that you can bring and ideally release to the public.
• describe more specific ideas within a subtheme
• propose new subthemes (please state "proposed by" and your name)

Ideas for themes

1. Spectral CT

2. Limited angle

3. Dynamic CT

https://fips.fi/dataset.php#stempo

4. PET/MR reconstruction incl. motion-compensation

5. Connect CIL to PyTorch

6. Connect CIL to ELSA projectors or algorithms

https://ciip.in.tum.de/elsadocs/

7. Connect to Helsinki data format

https://fips.fi/dataset.php
https://github.com/Diagonalizable/HelTomo
https://doi.org/10.5281/zenodo.7418878

8. Improve connection TIGRE

https://github.com/CERN/TIGRE

9. Improve connection to gVXR

https://gvirtualxray.fpvidal.net/
https://github.com/WebxCT/WebCT

10. New Micro-CT data readers, e.g. Tescan, YXLON, …

11. Nuisance parameters e.g. alignment parameters

12. SAR CPHD data

13. Non linear problems

https://github.com/SyneRBI/SIRF/issues/738
https://github.com/SyneRBI/SIRF/issues/739

14. Motion Compensated Cone-Beam CT Reconstruction

Existing collections of tomographic data

• https://zenodo.org/communities/ccpi/
• https://tomobank.readthedocs.io/en/latest/
• https://fips.fi/dataset.php
• https://wiki.cancerimagingarchive.net/pages/viewpage.action?pageId=52758026