Hi,

I general that looks OK to me.
Did you check that the registrations all look right for the principle images (rather than the ROIs).
That is, does the highres2standard_warp result produce a good version of the highres image that is aligned to the MNI?
Does the highres2standard_warp_inv produce a good version of the MNI image that is aligned with your highres (as well as your highres_2mm)?
If you haven't looked at these, test them by using applywarp with the appropriate image as the --in.

If these images do not look good then we need to address that first, independent of the ROIs.

If these images are all looking OK but the spherical ROIs are not, then send me the results of "fslhd" on a spherical ROI, the highres image and the highres_2mm image.  Also send me the results of "fslstats LmidoccSph -R -r".

All the best,
Mark


On 8 Sep 2012, at 14:01, "Mancini, Flavia" <[log in to unmask]> wrote:

Dear experts

I am created spherical ROIs and registered in functional space, with the intention of extract timeseries and use them as seed regions for PPI analysis.
I didn't have any problem creating the mask, but the registration of the mask is completely off in many subjects (mask appears flipped A-P, L-R, or without any systematic offset in some cases), therefore I would like to check that I haven't done any mistake in my analysis. Specifically:

1. I created a spherical roi (looks OK in fslview):
fslmaths $FSLDIR/data/standard/MNI152_T1_2mm_brain -mul 0 -add 1 -roi 57 1 15 1 36 1 0 1 Lmidocc -odt float
fslmaths Lmidocc -kernel sphere 5 -fmean LmidoccSph -odt float

2. In my first (non ppi) FEAT analysis I used nonlinear registration to standard space (12 DOF). 
Since the inv warp was taking to long I first created a 2mm highres warp:
flirt -in highres -applyisoxfm 2 -ref highres -out highres_2mm
I then created the inverse of highres2standard_warp with:
invwarp --force --ref==highres_2mm --warp=highres2standard_warp --out=highres2standard_warp_inv
(I had to force the command for the alarm message of too large ROI)

3. Finally I registered my mask in functional space:
applywarp --ref=example_func --in=LmidoccSph.nii.gz --warp=highres2standard_warp_inv --postmat=highres2example_func.mat --out=LmidoccSphFS

Many thanks,

best wishes

Flavia

---
Flavia Mancini, PhD
Research Associate
Institute of Cognitive Neuroscience
University College London