Hi,

You want to keep roughly the same FOV (with the same parts of the head)
that you can see in the MNI152 images.  If your protocol for placing
subjects in the scanner is quite consistent then you might be able to  
use
the same values for most subjects.  Otherwise you'll need to use  
fslview.
If you just go to the extreme coordinates of what you want in fslview  
then
the input to fslroi is the xmin, ymin, zmin from the smallest values and
xsize = xmax - xmin + 1 ; ysize = ymax - ymin + 1; and zsize = zmax -  
zmin + 1
Note that these are all voxel coordinates, not mm coordinates.

All the best,
	Mark


On 30 Oct 2008, at 03:54, Anthony Ang wrote:

> Hi Mark,
>
> Can you tell me what is an easy way to obtain the xmin, xsize, ymin,  
> ysize, zmin, zsize of the fslroi to reduce the FOV both for the  
> unstripped (_resize) and stripped (_n3ed) brain?
>
> For the stripped brain, I use fslview and decrease x, y, z  
> coordinates until the brain in the coronal, transverse and sagittal  
> view disappears to get xmin, ymin and zmin; then increase x, y, z  
> until the brain again disappears to get xmax, ymax and zmax.
>
> For the unstripped brain, it is not as easy to do this. What parts  
> of the skull should I retain in order for fnirt registration to work  
> efficiently?
> Is there any easier way to obtain the fslroi values for reducing the  
> FOV?
>
> Thanks.
>
> Best regards
> Anthony
>
>
> On Tue, Oct 21, 2008 at 2:10 AM, Mark Jenkinson  
> <[log in to unmask]> wrote:
>
> As for the invwarp, it does indeed take a *very* long
> time to run with your large FOV, as I suspected.
> With a smaller FOV I got it to run in just under 4 hours
> which is still quite slow, but more acceptable.  To do this
> just do the following rois:
>  fslroi x1000024_resize x1000024_roi 29 182 0 186 41 176
>  fslroi x1000024_n3ed x1000024_roi_n3ed 29 182 0 186 41 176
>
>