[ITK-users] deformable surface registration

[somi]

Hi,
For my project I ended up using a diffusion based algorithm somewhat
similar to described in
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.87.6924&rep=rep1&type=pdf

Here is a summary of the algorithm ( section 4 in paper above ) :

*A time and space discretized solution the the geometry-constrained
diffusion may be obtained by iterative Gaussian convolution and
back-projecting the constrained surfaces. The crux of the algorithm then
becomes : *
*1. Initial displacement. Construct an initial guess of the displacement
field. *
*2. Diffusion step. Convolve the displacement field with a Gaussian
kernel. *
*3. Deform source. Deform the source surface with respect to the
displacement field. *
*4. Matching (Projection onto the target surface). For all points on the
deformed surface: Find the closest point on the target surface. *
*5. Update displacement field. For all points on the deformed surface:
Change the displacements according to the match. *
*6. Convergence. Is the displacement field stable? If not, go to 2. Some of
the steps are explained in greater detail below*

I also used the features ( curvature in step 4) to determine the closeness.
The algorithm took a surface model and a point set as input. The vertices
of the surface model were deformed to match the point set without modifying
the connectivity.

Thanks,
Somi


Confessions, Obsessions , Beliefs , Wishes , and a few Esoteric Blogs
http://500px.com/seesomi

On Mon, Apr 27, 2015 at 12:32 PM, brian avants <stnava at gmail.com> wrote:

> hi michel
>
> nick tustison did some good work on this recently and most of it is
> available in both ITK and in ANTs. the work is evolving (in terms of
> efficient memory use) but the basics of doing images +
> segmentations/surfaces is available.  e.g. you could run a demons +
> iterative closest point registration pretty easily.
>
>
> brian
>
>
> On Mon, Apr 27, 2015 at 12:12 PM, Audette, Michel A. <maudette at odu.edu>
> wrote:
>
>>  Dear ITK users,
>>
>>
>>
>> one of my students needs to produce an accurate surface mesh
>> correspondence in conjunction with stats shape aware active surface models,
>> towards a vertebral segmentation application.
>>
>>
>>
>> Doing a Google search on this, I found a discussion from 2009
>> http://public.kitware.com/pipermail/insight-users/2009-August/031857.html;
>> I was wondering if there has been progress in this area since then. For
>> example, can the image deformation methods such as Demons and the like be
>> applied to a surface to surface correspondence problem (with further
>> information available such as underlying MRI gradients)?
>>
>>
>>
>> Our surface meshes relate to vertebral surfaces, so these are not
>> topologically equivalent to spheres; they are more like doughnuts.
>>
>>
>>
>> Thanks for kind support.
>>
>>
>>
>> Michel
>>
>>
>> Michel Audette, Ph.D.
>> Assistant Professor,
>> Department of Modeling, Simulation and Visualization Engineering,
>> Old Dominion University,
>> Norfolk, VA.
>>
>>
>> _____________________________________
>> 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.php
>>
>> 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://public.kitware.com/mailman/listinfo/insight-users
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/insight-users/attachments/20150427/ab9451fc/attachment.html>