Thank you for your clear and quick response. Much appreciated!
Chaleece
************************************************
Chaleece Sandberg, M.A.
Ph.D. Candidate
Speech Language and Hearing Sciences
Boston University Sargent College
Lab Manager
Aphasia Research Laboratory
On Fri, Jun 15, 2012 at 12:00 AM, MCLAREN, Donald
<[log in to unmask]> wrote:
>
> On Wed, Jun 13, 2012 at 11:28 AM, Chaleece Sandberg <[log in to unmask]> wrote:
>>
>> Hello!
>> I am wondering how SPM performs this function and what the data are that
>> I'm looking at. The process is to choose a peak voxel for a particular
>> contrast and then extract the eigenvariate for the area around that voxel.
>> When using a sphere, it makes sense that the peak voxel is at the center of
>> the sphere. Is it correct to assume that the eigenvariate is computed using
>> all voxels in that sphere? If so, then what happens when using a mask? Is
>> the eigenvariate computed using all voxels in that mask, or just those in
>> the mask that belong to the peak's cluster? Does it even matter which peak I
>> choose if I'm using a mask?
>
>
> It chooses all voxels that are above the threshold, in the mask, and inside
> the sphere. If a voxel doesn't meet all three criteria then its not
> included. Depending on the radius of the sphere, the peak voxel that you
> select may or may not make a difference.
>
>>
>> When it lists the number of voxels in the graphics window under the graph
>> of the eigenvariate, are those the voxels that are active at the chosen
>> threshold within the chosen mask? In other words, if there are just a
>> handful of voxels, the computation is from those voxels and not the whole
>> mask? So, for example, if there are 6 voxels in the RH mask, but 60 in the
>> LH mask (in areas that are roughly the same size bilaterally), is there a
>> built-in correction for the fact that one is much more "active" than the
>> other?
>
>
> The eigenvalues are computed for the active voxels. There is no correction
> for the number of voxels.
>
>>
>> I hope these questions make sense. I haven't found any other source for
>> answering them, so any insight would be greatly appreciated!
>> Thanks!
>
>
|