[Insight-users] best deformation registration algorithm?

[Tim Bhatnagar]

Hello,

I have been working with ITK very slowly over the past year. My goal is to
quantify tissue motion from MR images (the tissue experiences shearing and
compression). I have been working with an adaptation of Deformable
Registration 15 from the modules provided with ITK, but I am having
difficulty getting the results I'm looking for. Perhaps someone could
answer these questions I have (or provide a 'next step' suggestions?)

1) I am using MR images, same resolution, roughly same spatial size. There
is a relatively large amount of tissue movement in a localized area.
DefReg15 uses rigid->affine->coarse B-spline->fine b-spline, which
intuitively makes most sense for me based on my data. Is there any
suggestions that I should try a different algorithm?

2) My image voxel size is approx. 50x50x50 (but voxels are not isotropic).
Is there a 'rule of thumb' on choosing how many nodes the coarse and fine
b-spline registration steps have?

3) I am currently analyzing the output deformation fields in Paraview - is
there a convenient way to really get a measure of how the
algorithm-deformed data compares to the experimental data (ie using a
program for image comparison?)

4) This algorithm employs Mattes Mutual Information metric to evaluate the
registered image - should I be looking into others?

Thanks,

-- 
Tim Bhatnagar
PhD Candidate
Orthopaedic Injury Biomechanics Group
Department of Mechanical Engineering
University of British Columbia

Rm 5000 - 818 West 10th Ave.
Vancouver, BC
Canada
V5Z 1M9

Ph: (604) 675-8845
Fax: (604) 675-8820
Web: oibg.mech.ubc.ca
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20120813/0187cbfe/attachment.htm>