Steve,

That sounds like a reasonable request. Can you enter it into the bug tracker and assign it to me for now.

Bill

On Oct 31, 2007 6:24 PM, Steve Pieper <pieper@bwh.harvard.edu> wrote:

Here's a little more follow up on topic (1) from the original email.

It turns out that the output of the linear registration is always an
axial scan independent of the orientations of the input fixed and moving
scans (this is because both are changed to axial with an
OrientedImageFilter before the registration).

So really the request is that the registration routines should output
the transform and they should have options to specify the parameters of
the output image to either match one of the input images or to conform
to some settings on the command line (like output spacing, directions,
dimensions, etc).

-Steve

:
>      >>>>
>      >>>>     Hi -
>      >>>>
>      >>>>     2 Requests from our DBP:
>      >>>>
>      >>>>     1)  Sylvain has pointed to an issue with the way ITK
>     and/or the
>      >>>>     LinearRegistration command line module in Slicer3 handles
>      >>>> resampling the
>      >>>>     moving volume to match the fixed image.  Essentially the meta
>      >>>> data is
>      >>>>     correct (images load and display correctly in slicer) but the
>      >>>> pixels
>      >>>>     are
>      >>>>     not actually resampled the same.
>      >>>>
>      >>>>     The DidThis.txt file in the following directory shows the
>      >>>> difference.
>      >>>>     The input files are there too.
>      >>>>
>      >>>>     http://slicerl.bwh.harvard.edu/tmp/OrientationAndOrigin/
>      >>>>