Dear Klaas
Thanks for you reply.
It does look like the smoothed VOIs produce stronger connectivity
parametersin DCM. However, I don't know whether this is due to unwanted
overlap between VOIs, or simply because the smoothed data contains less
noise.
Can you please direct me to a formula to compute the degree of dependency
between the VOIs, based on the two Gaussians?
Thanks a lot
Tali
On Wed, 15 Jul 2009 09:40:26 +0000, Klaas Enno Stephan
<[log in to unmask]> wrote:
>Dear Tali,
Since you know the spatial distance between your peaks and the size of your
smoothing kernel, you could simply compute the degree of dependency in
terms of the overlap of the two Gaussians. Also, you could run the DCM on
the smoothed data first, and then apply it to the unsmoothed data, hopefully
producing parameter estimates that are not too dissimilar?
Best wishes,
Klaas
________________________________
Von: Tali Bitan <[log in to unmask]>
An: [log in to unmask]
Gesendet: Montag, den 13. Juli 2009, 11:58:58 Uhr
Betreff: [SPM] distance between VOIs in DCM
Dear DCM experts
We would like to use DCM with regions that are pretty close together (~16mm
between peaks at the group and individual level), while our smoothing kernel is
5mm.
Would it be reasonable to use 5mm radius VOIs based on these peaks
(resulting in a distance of only 6 mm between voxels in the perimeters of
these VOIs)?
Or is it better to use unsmoothed data? or alternatively - pick a different
(weaker) coordinate for one of the VOIs?
Thanks a lot
Tali Bitan
Haifa University
|