[Insight-users] help on registering two label volume (mask
volumes)
Luis Ibanez
luis.ibanez at kitware.com
Sat Oct 28 12:14:13 EDT 2006
Hi Alex,
0) There is no such thing as "the best filter" for registration.
The concept of "BEST" method is an intellectual abomination
disseminated by the decadence of the publishing system in our
community. The word "BEST" is what unscrupulous authors like
to write in the "Abstract" and "Conclusion" sections of their
paper in order to trick ignorant reviewers and naive readers.
Whoever claims to have a "BEST" method for doing something,
is *confused* or *trying to confuse you*.
1) You are right, ITK will take pixel spacing, and image origin
into account when doing registration.
2) Since you are registering labelelled images you probably want
to try the Cardinality Match image metric:
http://www.itk.org/Insight/Doxygen/html/classitk_1_1MatchCardinalityImageToImageMetric.html
3) If you want to do elastic registration (I assume that by "elasic"
you mean
"a deformation with more degrees of freedom than an affine"
you may want to try the BSplinesDeformableTransform
http://www.itk.org/Insight/Doxygen/html/classitk_1_1BSplineDeformableTransform.html
that is described in detail in the ITK Software Guide
http://www.itk.org/ItkSoftwareGuide.pdf
4) For Rigid registration you may want to try
the VersorRigid3DTransform:
http://www.itk.org/Insight/Doxygen/html/classitk_1_1VersorRigid3DTransform.html
5) If you want to do rigid plus scalilng then you may want to try
the Similarity Transform.
http://www.itk.org/Insight/Doxygen/html/classitk_1_1Similarity3DTransform.html
BUT
PLEASE
Before you start writing any code, read the Image Registration
chapter from the ITK Software Guide. From the description of
your problem, it is clear that you will benefit from spending
some quality time getting familiar with the concepts used in
the image registration framework implemented in ITK.
Regards,
Luis
---------------------
Alex Houston wrote:
> I am trying to register/allign two manually labelled volumes. Both have
> different spacing (in X and Y but same in Z) and different dimensions
> (in Z)...I know ITK does take of this things in registration but just
> want to make sure.
>
> Can anybody guide me what is best filter to use
>
> 1. I plan to do elastic registration
>
> 2. I plan to do rigid registration + scaling
>
> 3. Any particular metric or optimizer I need to use because I have
> labels (having values 1,2,3.4) in uint8 volume
>
> Thanks in advance
>
>
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
> <http://us.rd.yahoo.com/mail_us/taglines/postman8/*http://us.rd.yahoo.com/evt=39663/*http://voice.yahoo.com>
More information about the Insight-users
mailing list