Dear Arturo,
> I just tried to download this software and it told me the link no longer
> exists. Can you tell me where I can get it?
I think Karen mistyped the address slightly in her earlier message.
It is at:
ftp://ftp.physiol.ox.ac.uk/pub/matthew.brett/Corr_p
> Also, how do you generally go
> about determining the size of the volume of interest. It sounds like a
> dumb question but if you give me a simple example I'll extrapolate from
> there. Let's say I wanted to use dorsolateral prefrontal cortex as my
> volume of interest.
Well, the way that I have gone about it here, is to use an image editor, in
my case Analyze, and to go through the individual MNI brain image:
<spm>/canonical/T1.img
slice by slice, and edit out (set to 0) all pixels that are not in the
dorsolateral prefrontal cortex. This can generate some strange effects
in the z axis (assuming you were going through the image axially), where
you don't define the area in exactly the same way from slice to slice.
This needs to be sorted out, because it will falsely increase the surface
area of the region (see the README file for more details). I have, so far
therefore processed the edited image thus:
1) Binarised the image so that there are 1s in the region of interest and
zeros elsewhere. You can use the SPM ImCalc button to do this: press the
button, select the edited image, type "i1 ~= 0" (without the quotes) into
the formula box, and then type a name for the new binarised image e.g
bin_dlpfc.
2) Smoothed the image to 5mm to blur out the edges a bit -> sbin_dlpfc.img
3) Thresholded it to zero out the blurred edges of the image which are
far away from the region you defined. For example, ImCalc again, on the
sbin_dlpfc.img, with a formula of "i1 > 0.25" . It would be worth
checking your region to make sure it is roughly what you expected, and not
too large (it will likely be a little larger than you defined; try increasing
the threshold if this is a problem - e.g "i1 > 0.5".
The end result should be an image, which is non-zero in the area
corresponding to the DLPFC (in our case all the pixels = 1), and zero
outside the area.
You could then use this thresholded image as the image defining the
region of interest, in the vol_corr program.
Hope that helps, let me know if it's not clear,
Cheers,
Matthew
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|