sdcflows ds004024 misalignment of BOLD and fieldmap

What happened?

Checking w/ @jbwexler. Will record more details as I get them. Screenshot attached.

Possibly related: the fieldmap reportlet had weird coordinate listing

What command did you use?

tacc-openneuro submitter

What version of the software are you running?

fmriprep 23.2.0

How are you running this software?

Singularity

Is your data BIDS valid?

Yes

Are you reusing any previously computed results?

No

Please copy and paste any relevant log output.

No response

Additional information / screenshots

Jan 23 '24 21:01 effigies

@psadil Does this look like #461 to you?

Oct 04 '24 19:10 effigies

@jbwexler It's probably too much to hope you can find the dataset that this happened in?

Oct 04 '24 19:10 effigies

Unfortunately it's long gone as far as I can tell

Oct 04 '24 20:10 jbwexler

I could rerun it though if that would be helpful.

Oct 04 '24 21:10 jbwexler

I'm not sure. I haven't see doubling of the brain like that. But I have seen cases with fieldmaps that look like what's shown, by which I mean something like "squished into a line".

And the resulting SDC

Oct 04 '24 22:10 psadil

And FWIW, here's the fieldmap on that case Screenshot 2024-10-04 at 18 13 30

Oct 04 '24 22:10 psadil

I haven't see doubling of the brain like that.

I think the screenshot was just during the transition.

Oct 04 '24 23:10 effigies

I think the screenshot was just during the transition.

Oh, of course. Then, yes, this does seem related. I would be curious whether the coregistration works better with the dilated masks.

Oct 05 '24 02:10 psadil

One aspect of what I was seeing is that the step was highly variable (which I attributed to the sampling_percentage). If there's a way to regenerate the working directory, it could be interesting to see what repeated runs of that coregister node produce

Oct 05 '24 11:10 psadil