[Insight-users] 3D image registration

[Karthik Krishnan]

Regsitration in ITK consists of 4 components:
-metric
-transform
-optimizer
-Interpolator

Given that the transform is specified in physical co-ordinates ( 
accounts for pixel spacing ), you could try 3D registration on thick 
slice data as well. Be sure to use a decent interpolator. ( certainly 
not Nearest neighbor ).

The choice of the the transform is entirely up to your study depending 
on whether you anticipate purely rigid misalignments, pure translations, 
rotations + translations,  deformations. There are a variety of 
transforms for each of them.

If you are registering MR-MR images, you could use a MeanSquares metric 
coupled with a Gradient Descent optimzer. If you are registering MR-CT 
you could try one of the Mutual Information metrics. The GradientDescent 
optimizers are susceptible to noisy images.  You might consider 
smoothing the MR images prior to registration.

-karthik

Grace Chen wrote:

> <> Hi all,   I post my question a few days ago and didn't get any 
> feedback.  So, I post my question again in the next pargraph hoping 
> someone today will be able to help me with this problelm.  I am a new 
> ITK user, so my question should be trivial for those who use ITK's 
> registration tools.  Please help!! :>  Thanx!!   ----- Original 
> Message ----- *From:* Grace Chen <mailto:Grace.Chen at swri.ca> *To:* 
> insight-users at itk.org <mailto:insight-users at itk.org> *Sent:* Thursday, 
> September 01, 2005 12:00 PM *Subject:* [Insight-users] 3D image 
> registration
> Hi there,  Good day!   We are using ITK for registering 2D images for 
> our current project.  It's working very nicely for correcting most of 
> the misalignment in our data.  However, we noticed 
> out-of-plant movement in our image data, so we want to extend our 
> project to perform 3D rigid registration.  The data I have is the MR 
> scan for the brain.  The dimension of the image is (256, 256, 8), and 
> the spacing is (0.93mm, 0.93mm, 7mm).
> So, our problem is: since the space between neighbor slices is so 
> distant, can the ITK
> registration program perform 3D registration properly on this type of 
> data? ...or do we have to
> use some additional VTK/ITK tools to cope with that?   If we can use 
> ITK to register our data, can you recommand the registration component 
> good for our data?   Any help is greatly appreciated!! :>   Have a 
> great day!   Grace
>
> ------------------------------------------------------------------------
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
>
>------------------------------------------------------------------------
>
>_______________________________________________
>Insight-users mailing list
>Insight-users at itk.org
>http://www.itk.org/mailman/listinfo/insight-users
>  
>