I am trying to apply distortion correction using FEAT GUI on an fMRI data set. I have done this many times with very few problems. However, my last study used a tighter FOV. The whole brain was acquired but there was some aliasing along the phase encode direction for both the fMRI data and the Field Map. As a result of this aliasing I am having trouble using the FEAT GUI without some manual processing of the data.
Here is what I have tries so far
1) Remove the aliasing with fslroi and fslmerge from the fieldmap and fmri data set. This procedure has been described in the tutorials and on the discussion board. It seems to work well. It does modify the shift the qto_xyz:2 by 34.375, which equals to the 11 pixels I shifted times the pixel dimension of 3.125.
2) Skull strip the fieldmap (magnitude image) with bet2 to create fmap_mag_brain.nii.gz This works great and I haven't encountered any problems.
3) Skull strip the first volume of the fMRI data with bet2, create a mask, and then apply the mask to the 4D fmri data set.
4) Use FEAT GUI to setup my Prestats processing of the fMRI data. I have turned off Brain Extraction. BBR is selected - although it doesn't allow me to select another option.
5) Verify that all the images are in the same orientation in fslview. I also load the images into Freeview so I can see that the images overlay one another in physical space. Everything looks great. The fMRI data set has very little motion and images (except for the distortion) are almost perfectly registered.
6) Run feat from the command line. ( feat rfmri.fsf )
During the registration process the brain is shifted. This is most obvious in the fieldmap2edges.png. Where the field map and white matter edges have been shifted in the inferior to superior direction by almost 3/4 of the brain.
At this point, I have no idea what to do. Any suggestion or advice on debugging would be extremely helpful.
Thanks
Bob
|