I have absolutely no idea how best to deal with the resels per voxel data.
Even just achieving extremely accurate interpolation of T maps is not
easy, as they are the result of dividing one image by another (a
contrast image by a mean squared error). Contrast images can be
interpolated because they are just a linear combination of original
images. Mean squared error images can not be accurately interpolated.
Personally, I'd suggest warping either the original data or the
contrast images (depending on whether or not you are doing a second
level analysis). You'd do the smoothing after the warping.
Best regards,
-John
On 19 November 2011 10:31, Chris Filo Gorgolewski
<[log in to unmask]> wrote:
> Dear all,
>
> I want to use DARTEL to normalize T maps and later threshold them
> using topological FDR. The problem is that normalization will reslice
> the volume and the voxel sizes will change.
>
> spm_uc_clusterFDR as one of its arguments takes Voxels2Resels (V2R)
> which is basically 1/SPM.xVol.FWHM . I presume therefore that FWHM is
> measured in voxels? I so can I just adjust it by the voxel size ratio
> of volume before and after reslicing?
>
> Thanks in advance for any help.
>
> Best,
> Chris
>
|