Hi,
You can use voxel-wise EVs to remove the effects of a set of voxels for an individual subject. The EV needs to be zero for all good subjects and then have a value of one for the appropriate voxels in the appropriate subject. That will then let the GLM ignore the input from that subject in those voxels (other voxels, which contain useable data, should be zero in the EV). You may also need to fix the brain mask for the appropriate subjects so that this does not also remove the voxels from the analysis.
All the best,
Mark
> On 18 Mar 2017, at 18:26, Ishtiaq Mawla <[log in to unmask]> wrote:
>
> Hi all,
>
> I am wondering if at the group level, feat can deal with lower-level cope inputs that have zero values in some voxels/parts of the brain. this has happened due to loss of signal, not positioning FOV correctly to include the whole brain.
>
> I don't want to toss out full data from these subjects because they are missing certain voxels but I want to include them in the GLM analysis without biasing results by including zero values for some voxels.
>
> Theoretcally, there should be a way to calculate beta in each voxel only from subjects that contribute a non-zero value in a particular voxel, correct?
>
> Please advise, thanks!
>
> --Ishtiaq
|