ENH: Use *NiWorkflows*' EPI-reference workflow

[oesteban]

Replaces our workflows under dmriprep.workflows.dwi.util.

[josephmje]

Thanks for this @oesteban! It looks like it should also fix the issues I had with some oblique datasets. Is the old init_enhance_and_skullstrip_dwi_wf also meant to be replaced with init_brainextraction_wf from sdcflows?

[oesteban]

Is the old init_enhance_and_skullstrip_dwi_wf also meant to be replaced with init_brainextraction_wf from sdcflows?

I think we should give it a try, at the very least.

[josephmje]

This looks good to me! It's going to fail because the eddy nodes are no longer corrected properly. I can make a followup PR that includes the new masking workflow.

[oesteban]

This looks good to me! It's going to fail because the eddy nodes are no longer corrected properly.

I'll fix the connections tomorrow (my time). No need to break master for this.

[josephmje]

I'll fix the connections tomorrow (my time). No need to break master for this.

Thanks @oesteban. I checked out the reference epi and I don't think it looks like its supposed to.

[oesteban]

ugh - I had fixed that already. I'll double-check that the pinned versions of sdcflows and niworkflows are being used.

[josephmje]

This is looking better. The JHU1 epiref still looks a little weird but I tested it locally and it was fine there. For the masking reportlet, we are currently showing the unwarped epiref and mask. But should we instead show the epiref and mask before unwarping? The warped mask would be the input to head motion correction.

[pep8speaks]

Hello @oesteban, Thank you for updating!

Cheers! There are no style issues detected in this Pull Request. 🍻 To test for issues locally, pip install flake8 and then run flake8 dmriprep.

Comment last updated at 2021-05-18 21:35:00 UTC