The essential differences are twofold:
a) Smoothing is done before warping when you use (1), but is combined
with warping if you use (2). If you estimate native space contrast
images, and then spatially normalise these, then I would suggest not
smoothing the native data before fitting the GLM.
b) Masking is done differently in the two cases. If the 1st level
contrast images were generated without masking, and you supplied your
own mask for the second level analysis in both cases, then I would
expect the masking to be the same between the approaches.
Best regards,
-John
On 20 December 2011 10:19, Michel Dojat <[log in to unmask]> wrote:
> Dear All,
>
> For group analysis, two possible pipelines can be used when using Dartel for
> registration:
>
> 1) individual statistical analysis in individual referentials for all
> subjects.
> Then, applying Dartel deformation fields to the individual contrasts to be
> in the common referential (Dartel template) and then RFX analysis.
> In this case, the smoothing is performed in the individual referential
> before individual analysis.
>
> 2) application of individual Dartel deformation fields to individual
> functional images. Then individual statistical analysis in the common
> referential.
> and then RFX analysis. Here the smoothing is performed after deformation
> field application (and before individual statistical analysis).
>
> The results obtained are different.
> What is the best way to proceed and for which reasons ?
>
> Thank for your answer
>
> Michel
>
|