[Insight-users] NormalizedMutualInformationMetric

Luis Ibanez luis.ibanez at kitware.com
Mon May 12 11:18:27 EDT 2008


Hi Giorgos,


Thanks for providing this additional information.

The iterations do not look too bad...

The Scaling component of the Similarity transform
is particularly sensitive. What you may want to
consider is using a different parameter scaling for
this first element of the transform.

In fact, unless the scaling factor that you are
expecting is too large (e.d. x1.2 or 0.8x) then
you probably should solve first the rigid registration
of the two images, and only move into the Similarity
transform after you have an initial rigid registration.

Please try registering the two images first with just
a Rigid2D transform such as the CenteredRigid2DTransform
and let us know how it goes.


This will help to isolate the potential source of the
problem.


    Thanks


       Luis




  ------------------------
Giorgos Pichis wrote:
> 
> 
> 
> Hi Luis,
> 
> 
> 
>     -     BTW: What are the modalities of the images
> 
>               that you are trying to register ?
>     - I have to register CT images of bones . I am sending you two pairs
>     of them.
> 
>     -The terribleness of the results , has to do with the final
>     transform. It is way too far from 
> 
> the desirable .
> I am writing some of the iterations during the registration process.
> 
>     0 1.04009 [0.977812, 0.0250499, 97, 60, -0.0179536, 0.498556]
> 
> 10 1.03275 [0.925285, -0.0489282, 97.0129, 59.9924, 0.902668, 0.383388]
> 66 1.03271  [0.918458, -0.0767839, 97.0224, 59.9948, 0.941056, 0.343353]
> 105 1.03197  [0.915681. -0.0896959, 97.0199, 59.9825, 0.980653, 0.293535]
> 203  1.03138 [0.910645,  -0.110574, 97.0065, 59.9666, 1.05464, 0.161845]
> 303  1.03096  [0.904512, -0.130095, 96.9968, 59.9488, 1.15118, 0.0397982]
> 401  1.03023  [0.899097, -0.144235, 96.98, 59.9271, 1.16884, -0.120637]
> 498 1.02966  [0.894303,  -0.157513, 96.9616, 59.9319, 1.18778, -0.165064]
> 
> result:
> 500  1.02962  [0.894281, -0.15757, 96.9616, 59.9321, 1.18874, -0.165108]
>  
> I have to use Similarity2D Transform, because the images need scaling too.
> 
> 
> 
> 
> Thanks
> Giorgos
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> 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