Hello Jason,
If the data were not spatially smoothed, and one had a model for the spatial
signal (e.g., truncated Gaussian or spherical) then it would be possible to
estimate the number of voxels associated with each maximum via estimation of
the parameters of a spatial mixture model. But that would be quite a task,
frought with the assumptions about the spatial signal structure and probably
very unstable for closely spaced maxima. And even so, it would not be
possible to say that a given voxel "belonged" to one and only one maximum
because of the potential for overlapping spatial signals. If the data were
spatially smoothed, this latter problem would be worsened.
That is is my take on it anyway. Just a quick answer to tide you over until
a more substantial one comes along.
Eric
----- Original Message -----
From: "Chang, Jason (NIH/NINDS) [V]" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Friday, July 15, 2005 7:05 PM
Subject: [SPM] In clusters with mutiple maxima, figuring out how how many
voxels belong to each maxima
> To whom it may concern:
>
> Hello, I have just started using SPM2, and I have a question regarding the
> clusters and t-values that get displayed when one uses the "Results"
command
> to display the activation patterns.
>
> For instance, in one printout, one might get a cluster size (ke) of 10
> voxels with a corresponding t-value level of 5.0
>
> However, in other instances, one might get a cluster size (ke) of 100
voxels
> with multiple maxima--t-values of 4.0, 6.0, and 7.0. My question is as
> follows: is there a way to divide up the cluster so that you could
attribute
> a certain number of those 100 voxels to each t-value (i.e. find a way to
> divide up the voxels in the cluster so that you get something like 50
voxels
> belonging to the 7.0 t-value, 25 voxels belonging to the 6.0 t-value, and
25
> voxels belonging to the 4.0 t-value)?
>
> If this is possible and anyone knows how to display this type of result, I
> would appreciate it very much.
>
> Thank you very much,
> Jason
|