Dear Alessandro,
> When applying a small volume correction to a voxel/cluster within an a
> priori region of interest on an SPM created with a second-level random
> effects model, which search volume should be selected?
>
> For example, if I have a response in the amygdala on an SPM thresholded at P
> < 0.05 (uncorrected), should I select a "sphere" or "box" with the
> approximate dimensions of the amygdala or should I select "nearest
> cluster"?
For the analysis to be valid, you will need to specify the search region
independently of the analysis. In your case I suppose this means specifying
the sphere or box with the dimensions of the amygdala, and the centre in the
centre of the amygdala, rather than having the centre at your maximum voxel,
which would bias the analysis.
Best,
Matthew
|