FSL FEAT first level analysis with fmriprep output

Hi all,

After finishing preprocessing with fmriprep, I’m trying to use FSL FEAT for first level analysis, and get weird results in the stats folder under the .feat folder.

By “weird” I mean like this:

– fsleyes view of tsat3.nii.gz

I am using the 24 Head motion, the first 6 acomcor, and the cosine confounds as “additional confound EV’s”, under the “Stats” tab.

Tried smoothing FWHM with 4, and 5 mm, similar results.

I am using the space-MNI152NLin2009cAsym_desc-preproc_bold.nii.gz as 4D input

I tried registering it to MNI152-2mm following the Mumford solution of using FEAT on fmriprep output (How to use FEAT while skipping registration - YouTube)

Notice additional “weirdness” (roundness) of the design matrix:

– Design matrix, FWHM 5mm

In contrast with no smoothing at all:

Any clue what I’m missing?

I would appreciate your help with this, thank you.


Hi, I would like to admit my error - I supposed that the func file MNI152NLin2009cAsym_desc-preproc_bold output is skull stripped. It isn’t, and requires skull stripping.

Again, newbie at this