Dear Torsten,
I am fine with you using the method without the masks.
I simply wrote the last email for your information and to explain why
you got disastrous results when excluding all the background from
the mask.
All the best,
Mark
On 2 Aug 2011, at 15:57, Torsten Ruest wrote:
> thanks Mark! I got it working now with the masks weighting the missing bits with 0.1 and the common bits with 1 using a combination of intersection and union operations
> (fslmaths 103_retest-anatomical_brain_mask.nii.gz -mul func2highres_6dof_mask.nii.gz -mul 0.9 intersect
> fslmaths 103_retest-anatomical_brain_mask.nii.gz -add func2highres_6dof_mask.nii.gz -min 1 -mul 0.1 union
> fslmaths union.nii.gz -add intersect.nii.gz weighted).
>
> Now I am facing another problem:
>
> I got this weighting mask by registering the partial fMRI to the highres image using 6dof and then I generated a weighted mask using a union and intersection of masks obtained from the fmri and highres in highres space (see above). Unfortunately, as the images are not properly registered using 6dof and the fmri is a bit too much distorted, the weighted mask misses other bits which, so it seems, are crucial to a nice registration with 12dof. Hmm, I am quite happy with the results that I get with the normmi but without the masks, so unless you would oppose this procedure I would go ahead without the masks.
>
> Thanks very much for all your help, it's greatly appreciated!
>
> Torstn
>
|