[Insight-users] using VersorRigid3DTransform, failure to create output image? registration failure?

[Tim Bhatnagar]

Hello all,

I am using a somewhat modified (and I'm not entirely certain of the
modifications as I have inherited this program from a colleague) version of
VersorRigid3DTransform as a first step in a rigid-then-non-rigid
registration algorithm (the non-rigid portion is accomplished with 3D
B-spline deformable registration). The two image sets I have are MR slices
(48 slices per condition) of a live tissue before and after a relatively
large deformation. I've segmented only the tissue of interest from the
images.

After my first attempts with the 'Rigid' aspect of this algorithm, I am
receiving an output image with all pixel values of '0'. I do receive real
values for the translation vector, but my rotation matrix is essentially an
identity matrix (which could be feasible, as I'm not certain how much
rotation is going on during the deformation). Regardless, the aforementioned
colleague indicated that a 'zeroed' output image is signifying a
registration failure.

Can anyone give me some sort of indication of the limits of this rigid
registration program? If my deformation is too large or affine, should I be
using a different registration program before the 'non-rigid' part of the
algorithm?

My downfall is that I am really not well-versed in c++ or much of the
fundamental workings of ITK, but it seems to be a very powerful tool and
very much appropriate for my study. I am in the process of 'beefing-up' on
my understanding of the ITK environment, but in the meantime I would greatly
appreciate being able to move past this hurdle.

I can supply code if needed as well as any other pertinent info deemed
necessary. I am using the 3.20 ITK release.

Thanks so much in advance,

-- 
Tim Bhatnagar
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20110323/406ca24a/attachment.htm>