[Insight-users] motion correction results question

[Mark Wyszomierski]

Hi everyone,

I've been using some of the motion correction algorithms for some time with
contrast enhanced breast MRI cases. I'm specifically using registration
example 6 from build 2.0.0 (BSplineDeformableTransform) as an example. For
most cases it works very well. Every so often I do get some strange results.
I have included three links with small pictures showing the results of
registration of one image:

The stationary image:
http://www.mark-w.com/stationary.jpg

The moving image:
http://www.mark-w.com/moving.jpg

The result after registration:
http://www.mark-w.com/registered.jpg

I'm just wondering what could cause such a distorted result. Is it a
characteristic of the image itself that causes this? Is the change in
contrast from stationary to moving image so great that it's really getting
thrown off pretty badly? Is there a way I can limit the search space for
registration, so that the algorithm knows not to stretch the image so far?

Thanks for any info,
Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://public.kitware.com/pipermail/insight-users/attachments/20060206/d1e9510a/attachment.html