The normalise button often does not work so well, particularly for
fMRI that may have slightly different image contrast and artifacts
from the EPI templates. If you plan to normalise the EPI directly,
then I'd suggest using a segmentation approach to obtain the
appropriate deformations. This will be the default approach in SPM12,
and it appears to work reasonably well for a variety of different
image types.
Generally, the best way to figure out which of a number of approaches
is more accurate is to assess them empirically.
Best regards,
-John
On 26 October 2012 19:12, Niels Sigvartsen
<[log in to unmask]> wrote:
> Dear SPM experts,
>
>
>
> I’ve got a question about normalisation. When we normalise our EPIs with the
> parameters from segmenting the T1 (sagittal) structural, the results seem a
> smaller brain than if we normalise to the EPI template using the normalise
> algorithm.
>
>
>
> How can we assess which of the procedures might be better (and the one we
> want)?
>
>
>
> --
>
> Niels Petter Sigvartsen
>
>
>
> Research technician
>
> KG Jebsen Centre for Psychosis Research, TOP study
>
> Division of Mental Health and Addiction, Oslo University Hospital &
> Institute of Clinical Medicine, University of Oslo
>
>
>
>
|