hello Allison,
One common approach is to use only subjects for whom at least one voxel
activated at p<0.05 uncorrected can be found within a resonable distance
of your group peak or some a-priori landmark. DCM can then be performed
on a subset of subjects for whom reasonable activations were obtained.
However, if you would like to use the group ROIs, then the approach you
suggest of setting a threshold like p<1 (i.e. all voxels are
activated) will technically allow you to extract VOIs you want using the
SPM GUI.
good luck!
Chris
Christopher Summerfield
[log in to unmask]
On Tue, 11 Mar 2008, Allison Nugent wrote:
> Hello,
> I'm trying to run a DCM analysis on multiple subjects. After a
> traditional GLM group analysis, I selected several functional clusters, and
> defined ROI masks based on these group-level clusters. I then created new
> GLM models specific to my DCM design. I'm trying to use the VOI button to
> then extract the data in the masks I created at the group level. However,
> this seems to only select the suprathreshold voxels for whatever cluster and
> threshold I happen to be viewing at the time. What I would like to do is
> extract the data for all the voxels in the mask I defined at the group
> level. Is this possible? I know I can do it with Marsbar, but that won't
> give me the first eigenvariate. I can also just use an effects of interest
> contrast and set the threshold at 0.5 or something, but I was wondering if
> there was another option.
>
> Also, I'd like to know if there is a fix for the VOI display - it shows
> the black circle location of the VOI at whereever the cursur currently
> happens to be, and then appears to show a projection of the voxels onto that
> slice as red dots. Is there possibly a fix to make SPM display the VOI in
> the correct place for a mask image?
>
> Thanks!
>
> Allison
>
|