Hi Rodolphe,
There are several things to select activations. First, I would say the
z-score. However, as your result comes from a t or F statistics, all
cluster are valid. Of course you also have the extended threshold; a
high z-score for an isolated voxel does not seem to be pertinent.
Finally, it also depends on what you are interested in. SPM reports
cluster with peaks of activations and only presents distinct maxima for
a distance > 8mm (If I remember well). If you want to look at fine
spatial separation within a region it is better then to perform a ROI
analysis or something like that.
> Dear spmers,
>
>After a spm analyse you have a beautiful brainglass and the list of activations order by cluster z-score, coordinates, etc ...
>Sometimes, this list is very long. How do you decide wich activations you will keep as "significant" and which activations you wont notice. Do you proceed with z-score? (For exemple, taking all activations with a z>5 )
>
>Rodolphe N.
>Paris
>
>
Best,
Cyril
|