[Insight-users] Contour vs. Mesh registration
Eliana Vasquez Osorio
e.vasquezosorio at erasmusmc.nl
Mon Aug 17 12:45:03 EDT 2009
Hello Luis,
I missed the second part of your first e-mail. Would it be possible if
you point me to some info of registering 2D ultrasound images to 3D
pre-operative images?. May be I could use something like this, since
the surface actually comes from the delineation of the organ in CT
images, and the contour comes from an image of a cut of this organ
(slice)... So, in practice I could get the images.
Originally I was thinking on building a binary image from the contour
and match it to the surface using
ImageToSpatialObjectRegistrationMethod, but I'm not sure it will work.
And the second idea was: create a point set from the contour (dense
along the edges) and register it to a binary image (created from the
surface) using PointSetToImageRegistrationMethod, and once again, I'm
not sure it will work.
Do you think any of these ideas is plausible?
Thank you again,
Best regards, Eliana
Luis Ibanez wrote:
>
> Hi Eliana,
>
> Thanks for the clarifications.
>
> You are indeed dealing with a very hard problem.
> (and ill-posed one).
>
> I have no new suggestions, different from what I mentioned
> in my previous email.
>
> I could simply reinforce that you want to make use of any
> possible constrain you can add to this problem. (e.g. if you could
> determine the range of tilts that the plane may have).
>
> I know that people in this list has dealt with the similar problem
> of registering 2D ultrasound images to 3D pre-operative images.
> In that case, there is the advantage, of having some constrains
> on the orientation of the Ultra-sound plane, and on having access
> to the pixel values inside the object (organ), not just using values
> from the boundary.
>
>
>
> Regards,
>
>
> Luis
>
>
> -----------------------------
> On Mon, Aug 17, 2009 at 6:10 AM, Eliana Vasquez Osorio
> <e.vasquezosorio at erasmusmc.nl <mailto:e.vasquezosorio at erasmusmc.nl>>
> wrote:
>
> Hello Luis,
> Thank you for answering.
> It is a 2D manifold embedded in a 3D space. It's the surface of an
> organ (such as a kidney, liver, breast, etc...) and the contours are
> cuts on these organs.
>
> Any idea so far?
> Eliana
>
>
> Luis Ibanez wrote:
> >
> > Hi Eliana,
> >
> > Is the Mesh a 2D mesh ?
> >
> > That is, is embedded in a 2D space ?
> >
> > or is the Mesh a 2D manifold embedded in a 3D space ?
> >
> > (e.g. like the surface of a bone ?).
> >
> >
> > ---
> >
> > Unless you have other constrains that can help,...
> > this is a very hard problem.
> >
> >
> > One option that you may want to consider is to
> > compute the curvature of the planar curve (e.g.
> > the radius of the osculating circle). An then in
> > the Surface compute the min and max curvature
> > at every point.
> >
> > The curvatures of the planar curve (when matched
> > to the surface) will have to be inside the ranges of
> > the min and max curvatures of the matching points.
> >
> > Another way to look at it, is that the internal angles
> > of your polygon should have to be in the range of
> > the dihedral angles of consecutive faces of the Mesh.
> >
> > In that context, the easiest points to match will be the
> > points of the planar curve with the internal angles that
> > are farther from 180 degrees.
> >
> > ....
> >
> > In your email you ask if you could use Mesh to Image
> > registration methods.... Do you have the image from
> > which the Mesh was generated ?
> >
> >
> >
> > Please let us know,
> >
> >
> > Thanks
> >
> >
> > Luis
> >
> >
> > ------------------------------------------------------------------
> > On Wed, Aug 12, 2009 at 11:29 AM, Eliana Vasquez Osorio
> > <e.vasquezosorio at erasmusmc.nl
> <mailto:e.vasquezosorio at erasmusmc.nl>
> <mailto:e.vasquezosorio at erasmusmc.nl
> <mailto:e.vasquezosorio at erasmusmc.nl>>>
> > wrote:
> >
> > Hello itk-users,
> > I'd like to register a contour (polygon composed by coplanar
> > lines) to a
> > mesh (set of triangles floating in the space).
> > The contour comes from a cut of the mesh, but the location
> of this cut
> > is not known. The idea is to find the plane that was used
> to create
> > this cut. I understand that for some cases the result is
> not unique,
> > such as using a perfect circle vs. an sphere, but just one
> > solution is ok.
> > I wonder if someone has done this, or something similar, before.
> >
> > May be using ImageToSpatialObjectRegistrationMethod or
> > PointSetToImageRegistrationMethod ???
> >
> > Any idea/suggestion etc is welcome!
> >
> > Thanks, Eliana
> >
> >
> >
> >
> > --
> > Eliana M. Vásquez Osorio
> > PhD Student (OIO)
> > Erasmus Medical Center - Daniel den Hoed Cancer Center
> > Department of Radiation Oncology
> > Groene Hilledijk 301
> > 3075 EA Rotterdam
> > The Netherlands
> >
> > T: +31 10 70 41 491
> > F: +31 10 70 41 012
> > E-mail: e.vasquezosorio at erasmusmc.nl
> <mailto:e.vasquezosorio at erasmusmc.nl>
> > <mailto:e.vasquezosorio at erasmusmc.nl
> <mailto:e.vasquezosorio at erasmusmc.nl>>
> >
> > _____________________________________
> > Powered by www.kitware.com <http://www.kitware.com>
> <http://www.kitware.com>
> >
> > Visit other Kitware open-source projects at
> > http://www.kitware.com/opensource/opensource.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
> >
> >
>
>
> --
> Eliana M. Vásquez Osorio
> PhD Student (OIO)
> Erasmus Medical Center - Daniel den Hoed Cancer Center
> Department of Radiation Oncology
> Groene Hilledijk 301
> 3075 EA Rotterdam
> The Netherlands
>
> T: +31 10 70 41 491
> F: +31 10 70 41 012
> E-mail: e.vasquezosorio at erasmusmc.nl
> <mailto:e.vasquezosorio at erasmusmc.nl>
>
>
--
Eliana M. Vásquez Osorio
PhD Student (OIO)
Erasmus Medical Center - Daniel den Hoed Cancer Center
Department of Radiation Oncology
Groene Hilledijk 301
3075 EA Rotterdam
The Netherlands
T: +31 10 70 41 491
F: +31 10 70 41 012
E-mail: e.vasquezosorio at erasmusmc.nl
More information about the Insight-users
mailing list