ENH: Better integration of *SDCFlows*' unwarping #2576
Conversation
3d440fa to
ba62361
Compare
This PR builds on top of #2547, unifying the final BOLDref computation.
ba62361 to
16a833b
Compare
| @@ -524,6 +524,15 @@ def init_func_preproc_wf(bold_file, has_fieldmap=False): | |||
| niu.IdentityInterface(fields=["bold", "boldref", "mask"]), name="bold_final" | |||
| ) | |||
|
|
|||
| # Generate a final BOLD reference | |||
| # This BOLD references *does not use* single-band reference images. | |||
There was a problem hiding this comment.
Sorry, missed this in a previous PR. final_boldref_wf outputs to bold_final, which provides the reference image for bold_reg_wf. This is the primary use case for sbrefs, with the superior gray/white contrast. What are we now using sbrefs for?
There was a problem hiding this comment.
What are we now using sbrefs for?
TBH, I don't know. I've been meaning to address #1511 for a long while, and the SDCFlows integration has made the SBRef path more complicated.
That said, my impression is that #1511 is becoming really urgent after the integration. Maybe we should just push the priority of that issue up?
Yup - we were commenting right at the same time. See #2576 (comment) |
|
Okay, let's merge and address #1511 asap. |
This connection can (should) be shared by both the SE and the ME paths. That said, SDCFlows needs to be revised (nipreps/sdcflows#245) so that fieldmaps without HMC are generated (and remove that ``_pop`` from this connection). Ammends: #2576.
This PR builds on top of #2547, unifying the final BOLDref computation.
Changes proposed in this pull request
Documentation that should be reviewed