ENH: Precomputed derivatives #173
Conversation
Codecov Report
@@ Coverage Diff @@
## master #173 +/- ##
==========================================
- Coverage 32.54% 32.21% -0.34%
==========================================
Files 44 44
Lines 3650 3703 +53
==========================================
+ Hits 1188 1193 +5
- Misses 2462 2510 +48
Continue to review full report at Codecov.
|
| coregistration_wf = init_coregistration_wf( | ||
| omp_nthreads=omp_nthreads, | ||
| sloppy=sloppy, | ||
| debug="registration" in config.execution.debug, | ||
| ) | ||
| coreg_report_wf = init_coreg_report_wf( | ||
| output_dir=output_dir, | ||
| ) |
There was a problem hiding this comment.
Do we not still need these, even if there is a mask?
There was a problem hiding this comment.
since the T2w is only being used to generate the mask, nope. I was actually planning on removing the T2w restriction if a precomputed mask is available.
There was a problem hiding this comment.
Are we not outputting the coregistered T2w image? I think that's been a desired feature for fMRIPrep for a while, if it hasn't been explicitly brought up here.
|
This is ready for a final review - @hough129 @madisoth how has testing gone on your end? I'm working on a branch that implements |
|
I am still struggling to get it to fully complete - getting closer though. I believe @madisoth is having a separate issue that he reported on an issue thread. |
Are you running into an error?
Yes, though I believe its unrelated to this PR |
|
Yes - but I have a fix - the segmentation needs to be reproduced in order to test it. That will take a day or two. |
Co-authored-by: Chris Markiewicz <effigies@gmail.com>
this is the final straw
Not yet, but I'm hoping to beef up the test suite dramatically once this is in. |
Co-authored-by: Chris Markiewicz <effigies@gmail.com>
5 participants
Adds
--derivativesflag to allow for the use of precomputed derivatives (organized as one or more BIDS derivatives)Initial support is intended to allow the use of:
bold mask(Pushed to a later PR, as this will require bigger workflow refactor than the other two options)Addresses #166