Adds CIRIM

[wdika]

This pr adds the Cascades of Independently Recurrent Inference Machines (CIRIM), as presented in "Karkalousos, D. et al. (2021) ‘Assessment of Data Consistency through Cascades of Independently Recurrent Inference Machines for fast and robust accelerated MRI reconstruction’. Available at: https://arxiv.org/abs/2111.15498v1".

[codecov-commenter]

Codecov Report

Merging #156 (ca75ee1) into main (9e84f46) will increase coverage by 0.48%.
The diff coverage is 97.50%.

@@            Coverage Diff             @@
##             main     #156      +/-   ##
==========================================
+ Coverage   80.80%   81.29%   +0.48%     
==========================================
  Files          72       74       +2     
  Lines        5320     5480     +160     
==========================================
+ Hits         4299     4455     +156     
- Misses       1021     1025       +4     

Impacted Files	Coverage Δ
direct/nn/cirim/cirim_engine.py	95.45% <95.45%> (ø)
direct/nn/cirim/cirim.py	98.27% <98.27%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9e84f46...ca75ee1. Read the comment docs.

[jonasteuwen]

@wdika Can you rebase from main?

[wdika]

@wdika Can you rebase from main?

@jonasteuwen I rebased.

• The PR Labeler error is still here, due to privacy restrictions (?).
• Errors from codacy keep some kind of consistency on the repo, as they are the same on the other models and models' engines scripts. Not sure if I should fix them individually or that would go on a future request and a general refactoring for all models?

[georgeyiasemis]

A few comments

[georgeyiasemis]

@wdika Can you pull from main please? Black is failing because there is a new version

[wdika]

@wdika Can you pull from main please? Black is failing because there is a new version

Thanks @georgeyiasemis! I had to update Black, which now passes successfully.

[georgeyiasemis]

@wdika Codacy also complains about some issues: https://app.codacy.com/gh/NKI-AI/direct/pullRequest?prid=8627324

Can you take a look please?

[wdika]

@wdika Codacy also complains about some issues: https://app.codacy.com/gh/NKI-AI/direct/pullRequest?prid=8627324

Can you take a look please?

@georgeyiasemis thanks for the reply, I fixed the unused imports and variables errors. However, there are still four left about "Line too long" and "Too many local variables".

After checking the files of the rest implemented models I saw that those are common errors. So, could you please let me know if this would pass the standards now?

[jonasteuwen]

@wdika Can you please rebase from main and make sure all tests pass?

[wdika]

@wdika Can you please rebase from main and make sure all tests pass?

@jonasteuwen I rebased from main.

Still, the PR Labeler doesn't pass because of access restrictions. Codacy also complains but I guess those are in general "errors" of direct. Such as the "Line too long" when citing a paper and "Too many local variables" in the forward func of the model engine. As I see compared to the rest implementations those errors are present there as well.

Not sure if I need to change something there.