[Insight-users] ITK Image Registration

[Tom Vercauteren]

Hi Shoosh,

The best thing would be to read the paper that comes with the
submission. Theoretically, the diffeomorphic demons algorithm does
preserve the topology. Diffeomorphic mostly means invertible in this
setting. The algorithm looks for a deformation field whose Jacobian is
positive at each point. In practice though, in some hard cases,
numerical errors may lead to some small negative Jacobians.

If you need more control than invertibility (e.g. Jacobian > 0.1) then
you will need another algorithm.

Also my algorithm uses deformation fields defined at each pixel by an
independent vector. If you explicitly need B-Splines then you will
need another algorithm. (Note that the interpolation used with
deformation fields implicitly uses B-Splines, e.g. linear
interpolation is a spline of order 1).

The diffeomorphic demons code also works with a multi-resolution
approach but is not a multi-resolution B-spline basis.

Best regards,
Tom

2008/6/18 shoosh moosh <babygirl_25 at hotmail.com>:
>
> Does this type of registration preserve the topology of the images? I'm trying to model a deformation map as a hierarchical displacement field decomposed on a multi-resolution B-spline basis. The topology should be preserved by controlling the Jacobian of the transformation.
>
> ----------------------------------------
>> Date: Tue, 17 Jun 2008 17:28:00 +0200
>> From: tom.vercauteren at m4x.org
>> To: babygirl_25 at hotmail.com
>> Subject: Re: [Insight-users] ITK Image Registration
>> CC: insight-users at itk.org
>>
>> Hi Shoosh,
>>
>> I have submitted a diffeomorphic registration algorithm to the insight
>> journal last year:
>> http://hdl.handle.net/1926/510
>>
>> The code should be easy to integrate within an ITK development
>> environment and should follow ITK's features (cross-platform, N
>> dimensionnal, ...).
>>
>> If you find it useful, you're welcome to post a review :)
>>
>> You might also want to try it within a complete medical image
>> processing software. This algorithm is indeed integrated within
>> MedINRIA:
>> http://www.inria.fr/sophia/asclepios/software/MedINRIA/
>>
>> Regards,
>> Tom Vercauteren
>>
>> On Tue, Jun 17, 2008 at 4:39 PM, shoosh moosh  wrote:
>>>
>>>
>>> I'm trying to implement  3D image registration that preserves the topology of the images. Does anyone have any idea if this was implemented in ITK? If anyone has a link to the code please let me know.
>>> Thanks,
>>> _________________________________________________________________
>>> Discover the new Windows Vista
>>> http://search.msn.com/results.aspx?q=windows+vista&mkt=en-US&form=QBRE
>>> _______________________________________________
>>> Insight-users mailing list
>>> Insight-users at itk.org
>>> http://www.itk.org/mailman/listinfo/insight-users
>>>
>
> _________________________________________________________________
> Connect to the next generation of MSN Messenger
> http://imagine-msn.com/messenger/launch80/default.aspx?locale=en-us&source=wlmailtagline
>