Dear FSl experts,
a short update: the problem apparently lies with some (or rather each one) of the pre-processing steps.
I disabled all additional functions in Melodic and only ran the two registration steps and the resulting registration did not show the halo effect.
However, as soon as I enable a single one, or a combination of the:
- Mc Flirt Motion Correction
- Slice timing Correction
- BET extraction (on the fMRI, MPRAGE is run before)
The resulting registration has the same kind of error again.
I am confused how each one of these produces the same kind of error on their own.
As those don't have additional parameters to set (Or very limited choice, for slice timing I used the slice order I got told by the dcm2nii Converter, which is also Siemens standard), the default values are apparently not suited to my data.
However, as I have a sample of young, healty adults individuals without excessive motion (at least according to McFlirt and visual spot checks) all on the same scanner, I don't understand how for roughly half of them they work fine, and for the other half they create obvious errors.
I could do steps separately, but normally this should be things that work fairly well over a whole set of participants, shouldn't they?
I am really out of ideas what to try next, which is why I am hoping some of you can give me hints about what factors to look at.
All the best,
Judith Fronczek
|