Hi,

I'm not quite sure in what way this is failing, so it is difficult to  
advise you.
If the forward transform works well, then getting the inverse is a very
simple operation and nothing should go wrong.

Is the mask in the wrong place?
Is the mask too small or disconnected?

The likely problems are that you have not generated a floating point
output mask and then rethresholded it appropriately (see the FSL FAQ).
Alternatively, you may not be using the correct reference image for this
step.  Is it essential that you select the image from the target space  
(in
your case the 2x2x2 FA map) as the reference image.  It is also possible
that you have not generated the inverse correctly.  The easiest way
to generate the inverse is to use the InvertXFM gui.  The easiest way
to apply the transformation is with the ApplyXFM gui, but remember
that you need to threshold and binarise the output image.

I hope this solves the problem.
If not, please email us with more details.

All the best,
	Mark


On 8 Jun 2008, at 01:05, Anastasia Ford wrote:

> Hi,
>
> I am having problem registering a mask using inverse transform  
> generated  by FLIRT.
> I  registered an FA map to FMRIB58_FA_1mm and drew a mask of the  
> internal capsule on the
> registered FA map. I am trying to register this mask to the original  
> FA map using the inverse
> transform option of the FLIRT module. Unfortunately, the mask does  
> not register properly to the
> original 2x2x2 FA map. Could you recommend another way to perform  
> this registration/advise on
> the causes of the inverse transform registration failure?
>
> Thank you,
>
> Anastasia
>