Skip to content

Conversation

@joshuacwnewton
Copy link
Member

In PR #25 (commit 4893e6f), we switched to using the T2 initwarps. But, we haven't done that for the tutorials yet.

joshuacwnewton added a commit to spinalcordtoolbox/spinalcordtoolbox that referenced this pull request Mar 10, 2025
## Description

When the contrast-agnostic model was changed from MONAI to nnUNet, the
method for determining the output suffix also changed, thus the output
switched from `_seg.nii.gz` to `_sc_seg.nii.gz`. I think adding the "sc"
part makes sense for multiclass models (SC/lesion, SC/GM, etc.). But,
for our single-class SC tools (`sct_deepseg_sc`, `sct_propseg`, etc.) we
typically just output `_seg.nii.gz`.

So, this PR adds a special case to check for single-output SC models,
then choose the "basic" suffix accordingly.

This PR is being tested in
https://github.com/spinalcordtoolbox/sct_tutorial_data/pull/27/files#r1985521419,
because that PR is getting blocked by this issue (SCT course was written
for `_seg.nii.gz suffix`).

## Linked issues

Fixes #4805.
Needed for
spinalcordtoolbox/sct_tutorial_data#27, which is
in turn needed for
#4804.

---------

Co-authored-by: Mathieu Guay-Paquet <[email protected]>
@joshuacwnewton joshuacwnewton merged commit 0d1259a into master Mar 10, 2025
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant