[Insight-users] Deformable Registration

Luis Ibanez luis.ibanez at kitware.com
Fri Dec 17 07:56:06 EST 2004


Hi Kathrin,

Is the point on the ellipse really associated
to a landmark that you can see in the image ?

For example, is it a feature on the border of
the ellipse that has an intensity different
from the surroundings and it is recognizable ?

It will help a lot if you post a screenshot
of your ellipse so we can take a look at the
image.

If the point that you are trying to track is
a 'geometrical abstraction' and not a feature
in the image, then it is unlikely that you will
be able to track it from one image to another.

Again,... we will need to look at one of your
images in order to give you sound advice.


    Regards,


       Luis



-----------------
kathrin wrote:

> Hello,
> 
> thanks for your information!
> I've already read the description of the registration algorithms in the 
> itk-Software Guide. But the Deformable
> Registration in ITK just computes a deformation component for each 
> dimension. So the registration tells me,
> how the ellipse is deformed in each dimension.
> For my problem that is not enough!
> I'd like to compute the motion of one point between the two images, the 
> point is situated on the ellipse contour.
> I dont want to compute a general deformation between two contours. It is 
> possible that the ellipse contour
> rotates around the centroid and I'd like to get an individual rotation 
> angle for each point on the contour. Can I get this
> detailed information for several points with an registration algorithm??
> To recognize the ellipse is no problem.
> 
> Thanks,
> Kathrin
> 
> Luis Ibanez schrieb:
> 
>>
>> Hi Kathrin,
>>
>>
>> From your description, it seems to be reasonable to
>> solve this problem by using Deformable Registration.
>>
>> However, depending on the characteristics of your images,
>> this may end up being a trivial process for less than
>> an hour or...  a Ph.D. dissertation
>>
>> Note that deformable registration is a computationally
>> intensive task. It may be an overkill for the problem
>> that you are trying to solve. In particular in 3D.
>>
>> You may want to consider fitting a geometrical model to
>> the ellipse in one image, and then fitting a similar model
>> to the other image.  This can be done with the
>>
>>         SpatialObjectToImageRegistration
>>
>> You will find a description of this concept on the
>> ITK Software Guide
>>
>>
>>     http://www.itk.org/ItkSoftwareGuide.pdf
>>
>>
>> Section 8.14 "Model based registration", pdf-page 328.
>>
>>
>> If you want to give it a try to the DeformableRegistration
>> algorithms, probably BSplines is the one that will give
>> you a good balance between flexibility and computational
>> time. You will find examples for all the Deformable Registration
>> algorithms in ITK under the directory:
>>
>>
>>            Insight/Examples/Registration
>>
>>
>>
>>
>> Are the ellipses in the 3D images easily recognizable ?
>>
>> or are they immersed in a strong level of noise ?
>>
>>
>>
>>
>> Please let us know,
>>
>>
>>    Thanks
>>
>>
>>       Luis
>>
>>
>>
>> ----------------
>> kathrin wrote:
>>
>>> Hello,
>>>
>>> I wonder if the following problem can be solved with deformable 
>>> registration:
>>> I'd like to choose a point in a 3D image and I want the registration 
>>> algorithm to find the position of the
>>> same point in an other 3D image. The point is part of an ellipse, so 
>>> it's no vertex.
>>> The two images illustrate the same object at two different points of 
>>> time. During the period of time, the
>>> shape of the objects warps.
>>> Is it possible to solve this point corresponding problem with 
>>> deformable registration? If so, can you
>>> tell me, which registration algorithm to use?
>>>
>>> Thanks,
>>> Kathrin
>>> _______________________________________________
>>> Insight-users mailing list
>>> Insight-users at itk.org
>>> http://www.itk.org/mailman/listinfo/insight-users
>>>
>>>
>>
>>
>>
>>
>>
> 
> 
> 






More information about the Insight-users mailing list