search

nilearn / nilearn

Machine learning for NeuroImaging in Python
http://nilearn.github.io
Other
1.16k stars 588 forks source link

[ENH] better error when mask with unmatched affine/shape occur in `intersect_masks` #3740

Open htwangtw opened 1 year ago

htwangtw commented 1 year ago

Is there an existing issue for this?

Describe your proposed enhancement in detail.

I encountered this when creating a group epi mask for a dataset, all outputs are in native epi space from fMRIPrep.

Currently the function would raise an error when detecting a mask with unmatched affine/shape from the reference, but no further details.

This is useful, but I will need to almost duplicate the code of this function if they want to catch which image is causing the problem, or find out what's the reference/problematic affine/shape for further diagnosis.

I consider to refactor the code, mostly to separate the sanity check from the mask merging function in the current function. We can just raise the error with more details, and that will not change the input/ouput spec of the current function.

Benefits to the change

Clearer issue with what's going on when encountering an shape / affine mismatch.

Pseudocode for the new behavior, if applicable

No response

bthirion commented 1 year ago

Obviously this would improve UX and is in the scope of Nilearn. It's not very clear how much additional code you had mind, should not be much I guess.

htwangtw commented 1 year ago

I haven't started working on the code side yet, but can work on a PR and use that as a base of the discussion.