Hi Rasmus,

thanks for your comments.

Rasmus Birn wrote:

> Depending on the computations you intend to perform, the ROI
> selection could be performed "on the fly" using the i,j, and k
> variables (or x,y,and z variables) in 3dcalc.

Currently I use "on the fly", which works well but is not optimal in terms of speed.

> This way you
> would only have to store the i,j,k (or x,y,z) values for each
> ROI center and loop through those.

If we would only store the centers, that would pretty much amount to a conjunction mask of the skull-stripped brain and the EPI volume, wouldn't it?

Also please bear in mind that I would like to store more sophisticated ROIs than just spheres.

cheers,
Nick