[Insight-users] MultiResolutionPDEDeformableRegistration warping

[Michael G]

Hi Lydia,

I upgraded to v1.6.0 then copied the changes in CVS that you pointed out. 
The results were identical to when I used the old copies that came with 
v1.4.0. Running the DemonsRegistrationFilter
works so well, I'm not sure what could be distorting the results from the 
MultiResolutionPDEDeformableRegistration this much. I copied the steps I 
take before running the method, they are almost identical to the setup of of 
the DemonsRegistrationFilter:

1) Get image buffers from fixed / moving image
2) Run the moving image buffer through histogram matching image filter using 
fixed image as
    reference.
3) Set number of levels of registration method to 5
4) Set first # of iterations to 128 - thereafter each step has half the # of 
iterations as the last
5) Use warp image filter and the output deformation field to map the moving 
image back to the
    fixed image.
6) Write results to file.

You can really see severe distortions in parts of the image after 
registration is complete. Does my methodology sound correct? Thanks for the 
help!

Regards,
Michael

>From: "Lydia Ng" <lng at insightful.com>
>To: "Michael G" <michaelguiliametti at hotmail.com>
>Subject: RE: [Insight-users] MultiResolutionPDEDeformableRegistration 
>warping
>Date: Thu, 11 Mar 2004 21:40:25 -0800
>
>Hi Michael,
>
>A fix was made fairly recently that might be related to your problem.
>
>The relevant changes are in
>itkDemonsRegistrationFunction.h  revision 1.24
>http://www.itk.org/cgi-bin/viewcvs.cgi/Code/Algorithms/itkDemonsRegistrationF
>unction.h?cvsroot=Insight
>
>itkDemonsRegistrationFunction.txx revisions 1.22 and 1.23
>http://www.itk.org/cgi-bin/viewcvs.cgi/Code/Algorithms/itkDemonsRegistrationF
>unction.txx?cvsroot=Insight
>
>Do you want to have a go at figuring out the difference? Else it will take 
>me
>a few days to work out a patch.
>
>- Lydia
>
> > -----Original Message-----
> > From: Michael G [mailto:michaelguiliametti at hotmail.com]
> > Sent: Thursday, March 11, 2004 9:13 PM
> > To: Lydia Ng
> > Cc: insight-users at itk.org
> > Subject: RE: [Insight-users] MultiResolutionPDEDeformableRegistration
> > warping
> >
> > Hi Lydia,
> >
> > I'm still using v1.4.0,
> >
> > Regards,
> > Michael
> >
> >
> > >From: "Lydia Ng" <lng at insightful.com>
> > >To: "Michael G" <michaelguiliametti at hotmail.com>
> > >Subject: RE: [Insight-users] MultiResolutionPDEDeformableRegistration
> > >warping
> > >Date: Thu, 11 Mar 2004 21:03:53 -0800
> > >
> > >Hi Michael,
> > >
> > >Which version of ITK are using?
> > >
> > >-Lydia
> > >
> > > > -----Original Message-----
> > > > From: Michael G [mailto:michaelguiliametti at hotmail.com]
> > > > Sent: Thursday, March 11, 2004 4:23 PM
> > > > To: insight-users at itk.org
> > > > Subject: [Insight-users] MultiResolutionPDEDeformableRegistration
> > >warping
> > > >
> > > > Hello all,
> > > >
> > > > I've been working with the MultiResolutionPDEDeformableRegistration
> > > > registration method and am getting odd results. Initially I was 
>using
> > >the
> > > > DemonsRegistrationFilter method because I was getting great results
> > with
> > > > it,
> > > > but it was taking about 11 seconds a slice to register and was 
>hoping
> > >the
> > > > MultiResolutionPDEDeformableRegistration would speed things up. The
> > > > doxygen
> > > > notes say that MultiResolutionPDEDeformableRegistration uses the
> > > > DemonsRegistrationFilter as its internal registration method by
> > default.
> > > > But
> > > > I am getting some serious warping around the edges of my output 
>images
> > > > after
> > > > registration is complete. Are there any examples of using
> > > > MultiResolutionPDEDeformableRegistration that I can look at? What
> > > > parameters
> > > > could I be fouling up to create such intense distortion? I've
> > normalized
> > > > the
> > > > input images in the same way I would before handing them to the
> > > > DemonsRegistrationFilter method.
> > > >
> > > > My goal is to register breast MRI images with a contrast agent
> > injection
> > >-
> > > > is there a more optimal registration method to use for this that ITK
> > > > offers
> > > > besides the DemonsRegistrationFilter or
> > > > MultiResolutionPDEDeformableRegistration?
> > > >
> > > > Thank you,
> > > > Michael
> > > >
> > > > _________________________________________________________________
> > > > Store more e-mails with MSN Hotmail Extra Storage - 4 plans to 
>choose
> > > > from!
> > > > http://click.atdmt.com/AVE/go/onm00200362ave/direct/01/
> > > >
> > > > _______________________________________________
> > > > Insight-users mailing list
> > > > Insight-users at itk.org
> > > > http://www.itk.org/mailman/listinfo/insight-users
> >
> > _________________________________________________________________
> > Frustrated with dial-up? Lightning-fast Internet access for as low as
> > $29.95/month. http://click.atdmt.com/AVE/go/onm00200360ave/direct/01/
>

_________________________________________________________________
Create a Job Alert on MSN Careers and enter for a chance to win $1000! 
http://msn.careerbuilder.com/promo/kaday.htm?siteid=CBMSN_1K&sc_extcmp=JS_JASweep_MSNHotm2