Hi,
I think there could be two things that cause this:
1 - SPM can put NaNs into some voxels (a code for Not a Number) and FSL cannot cope with this. If this is the case then they need to be removed with fslmaths before any other processing in FSL.
2 - We recommend segmenting in the original structural space, and not a resampled space, because the latter has extra interpolation artefact and partial voluming that was not present in the original space. So if your normalising in SPM is also applying a transformation to standard space, then we would recommend against that.
All the best,
Mark
On 14 Jun 2013, at 07:08, SUBSCRIBE FSL Wily <[log in to unmask]> wrote:
> Hi expert,
> If I use bet,first to segment the hippocampus,I can get the right result.But if I use spm to normolise the sMRI,sometimes I am success,but almost time I am failure.I want to know the reasons.
>
>
>
>
>
>
> Thanks in advance,
> Wily
|