In many of our recent scans, we have started seeing some new problems in first-level analyses. It's a bit hard to describe/specify the problem, which is why we're taking it here to brainstorm.
1) Results are flagrantly wrong: we don't see functional activity in motor cortex for hand movements. This activity is normally quite apparent even at the single-run level. This is too experiment-specific for the list to troubleshoot, but it may be indicative of some other, deeper problem...
2) We've double-checked the EVs and they are fine.
3) We've double-checked the design files and they are unchanged from past design files (in fact, same design files we've used previously, with subject-numbers replaced.)
4) Something seems to be odd with the registration. Unusual shapes and orientations in report_reg. Example_func shows a lot of non-brain structures, including some stripey messy stuff down at the base of the brain (most easily visible in first slice of any rendered_thresh_zstat#.png file)
BUT: 5) The brain extraction looks pretty good, and I've run fsl_reorient2std on the whole-head and brain-extracted anatomicals.
Entire first-level results (500 MB) available here:
https://www.dropbox.com/sh/zt4q66v2u3emph6/AADp8DN4Z_UQ54GUiGvrV4zva?dl=0
Inputs (anatomical whole-head and brain-extracted, nifti inputs, design file, etc; 40MB) available here:
https://www.dropbox.com/sh/s6jm3llz1hu8h66/AABL5GvS8gXpSCQ5enf_b6oDa?dl=0
Thanks,
Benjamin Philip
[log in to unmask]
|