Jeff,
if you enter an explicit analysis mask in SPM (if I remember correctly
its not a
GUI option for fMRI modeling in SPM2, but you can do it by modifying
SPM.xM.VM), it can be any resolution. SPM will then only include functional
voxels which are in your mask using nearest-neighbour interpolation of your
mask data. In most cases, this is sufficient.
If you need a higher resolution, then you will have to do the analysis on
upsampled fMRI data.
Quoting Jeff Browndyke <[log in to unmask]>:
> Group,
>
> Currently, we have subject anatomicals (.93 x .93 x 1.3) co-registered to
> their functionals (3.75 isomorphic) in SPM2, and for reasons limited by our
> data, we are opting not to leave native space. The subjects' anatomicals
> were used to establish ROI masks in native space at the anatomical
> resolution, but at co-registration the functionals were not resliced to the
> anatomical voxel size to greatly conserve disk space (i.e., functional
> volumes per subject = 1002). How do we apply the anatomical ROIs traced at
> ~ 1 mm to the 3.75 mm functional space without resulting in a change in the
> binary nature of the masks (0/1) or distortion in the ROIs?
>
> If we need to reslice the functional data so-be-it, but I'm not looking
> forward to 8 GB per subject disk crunch and there's no guarantee that our
> systems will be able to handle analyses with such file sizes.
>
> Regards to all,
> Jeff
>
>
>
>
>
>
>
|