[Insight-users] different variations of demons registration

[Tom Vercauteren]

Hi Zoltan,

Yes, there might be some differences related to border effects,
smoothing steps, rounding issues and so on. See e.g.
http://www.itk.org/Bug/view.php?id=7351
http://www.itk.org/Bug/view.php?id=3973

My experience would favor the use of
FastSymmetricForcesDemonsRegistrationFilter vs
SymmetricForcesDemonsRegistrationFilter since I tried to make it as
stable as possible.

As a side note you might want to try the command-line tools provided in either
http://www.insight-journal.org/browse/publication/644
or
http://www.nitrc.org/projects/brainsdemonwarp/

They both provide easy means of comparing many demons versions.

Tom

On Thu, Feb 18, 2010 at 09:56, Zoltan Seress <seress.zoltan at gmail.com> wrote:
> Hi,
> Thanks for the answers. Then is there any reason to
> use SymmetricForcesDemonsRegistrationFilter
> if FastSymmetricForcesDemonsRegistrationFilter is at hand? Could be any
> difference between the registration results of them?
>
> On Thu, Feb 18, 2010 at 8:37 AM, Tom Vercauteren <tom.vercauteren at gmail.com>
> wrote:
>>
>> Hi Zoltan,
>>
>> FastSymmetricForcesDemonsRegistrationFilter and
>> SymmetricForcesDemonsRegistrationFilter use the same algorithm from
>> Thirion. The only difference lies in the implementation. As
>> illustrated by the name of the class,
>> FastSymmetricForcesDemonsRegistrationFilter  should just be faster...
>>
>> Tom
>>
>> On Thu, Feb 18, 2010 at 00:59, Luis Ibanez <luis.ibanez at kitware.com>
>> wrote:
>> > Hi Zoltan,
>> >
>> >
>> > It seems that you missed the text from the header file:
>> >
>> >      itkSymmetricForcesDemonsRegistrationFilter.h
>> >
>> >
>> > <quote>
>> > * This class was contributed by Corinne Mattmann, ETH Zurich,
>> > Switzerland.
>> >  * based on a variation of the DemonsRegistrationFilter. The basic
>> > modification
>> >  * is to use equation (5) from Thirion's paper along with the
>> > modification for
>> >  * avoiding large deformations when gradients have small values.
>> >  *
>> >  * SymmetricForcesDemonsRegistrationFilter implements the demons
>> > deformable algorithm that
>> >  * register two images by computing the deformation field which will map
>> > a
>> >  * moving image onto a fixed image.
>> > </quote>
>> >
>> >
>> > Is this the information that you are looking for ?
>> >
>> >
>> > The difference in computation time is due to the fact that this
>> > class computes gradients of both the Fixed image and the
>> > Moving image.
>> >
>> >
>> >
>> >      Regards,
>> >
>> >
>> >             Luis
>> >
>> >
>> >
>> >
>> > --------------------------------------------------------------------------------------
>> > On Wed, Feb 17, 2010 at 2:15 PM, Zoltan Seress <seress.zoltan at gmail.com>
>> > wrote:
>> >>
>> >> Hi all,
>> >> I have browsed the classes of itk 3.16 to get some information about
>> >> the
>> >> variants of the demons registration. I have found these:
>> >> 1, itkDemonsRegistrationFilter: this is the one algorithm by
>> >> Thirion: "Image
>> >> matching as a Diffusion Process: and Analogy with Maxwell's Demons"
>> >> 2, itkPointSetToSpatialObjectDemonsRegistration: this is another
>> >> algorithm
>> >> by Thirion
>> >> 3, itkDiffeomorphicDemonsRegistrationFilter: it is presented by
>> >> Vercauteren
>> >> (http://hdl.handle.net/1926/510) Algorithm 3 should be this one.
>> >> 4, itkFastSymmetricForcesDemonsRegistrationFilter: it is also presented
>> >> by
>> >> Vercauteren (http://hdl.handle.net/1926/510) Algorithm 2 should be this
>> >> one.
>> >> 5, itkSymmetricForcesDemonsRegistrationFilter: There is no additional
>> >> info
>> >> in the description. Has not been this implementation published
>> >> somewhere?
>> >> All I know that it uses both the moving image gradient and the fixed
>> >> image
>> >> gradient, and it is approximately 3 times slower
>> >> than itkDemonsRegistrationFilter. Can you tell me more about this
>> >> filter?
>> >> PS: I know about other variants such as log-domain diffeomorphic demons
>> >> and
>> >> symmetric log-domain diffeomorphic demons, but these are too much for
>> >> once.
>> >> --
>> >> Zoli
>> >>
>> >> _____________________________________
>> >> Powered by www.kitware.com
>> >>
>> >> Visit other Kitware open-source projects at
>> >> http://www.kitware.com/opensource/opensource.html
>> >>
>> >> Kitware offers ITK Training Courses, for more information visit:
>> >> http://www.kitware.com/products/protraining.html
>> >>
>> >> Please keep messages on-topic and check the ITK FAQ at:
>> >> http://www.itk.org/Wiki/ITK_FAQ
>> >>
>> >> Follow this link to subscribe/unsubscribe:
>> >> http://www.itk.org/mailman/listinfo/insight-users
>> >>
>> >>
>> > _____________________________________
>> > Powered by www.kitware.com
>> >
>> > Visit other Kitware open-source projects at
>> > http://www.kitware.com/opensource/opensource.html
>> >
>> > Kitware offers ITK Training Courses, for more information visit:
>> > http://www.kitware.com/products/protraining.html
>> >
>> > Please keep messages on-topic and check the ITK FAQ at:
>> > http://www.itk.org/Wiki/ITK_FAQ
>> >
>> > Follow this link to subscribe/unsubscribe:
>> > http://www.itk.org/mailman/listinfo/insight-users
>> >
>
>
>
> --
> Zoli
>