[Insight-users] Masks in ImageToImageMetric

[Luis Ibanez]

Hi Francois,


Your observation is correct. The MattesMutualInformation
implementation in ITK was not yet taking the Masks into
account.  The set of random pixels was being sampled from
the region that the user provides as the FixedImageRegion.

These points were mapped through the transform into the
MovingImage space and if they fall inside the physical
extent of the moving image, then they were taken into
account for the computation of the joint histogram.

Note that you could use the "FixedImageRegion" in order
to restrict the computation of the Metric to a rectangular
region of the fixed image.

---

This issue has been entered as a Feature Request in the
bug tracker.

http://www.itk.org/Bug/bug.php?op=show&bugid=817&pos=0


and...     has been implemented today.


Please update your CVS checkout in order to get this
modification.

You can now create a SpatialObject over your Fixed
image and another SpatialObject over your Moving
image in order to specify the region of pixels that
you want to participate into the Metric computation.


Please let us know if you have any further questions.



   Regards,


      Luis


--------------


Just for the record:

    The paper in which this variant of
    Mutual Information was published is :

"PET-CT Image Registration in the Chest Using Free-Form Deformations"
by Mattes, Haynor, Vesselle, Lewellen and Eubank


     Received  :       June    11  2001
     Revised   :       July    16  2002
     Published :       January     2003


     Total time:          18 Months

     Source code available with the paper:     No
     Image  data available with the paper:     No
     Research supported with public funds:    Yes


--------------------------------------------------
--------------------------------------------------
Bignalet Francois wrote:

> Hi all!
> 
> In itkImageToImageMetric there are 2 masks for the fixed and the moving image. In itkMeanSquareImageToImageMetric , in the method GetValue() it is always checked that the current pixel is in the mask, but it seems like this is not the case in itkMattesMutualInformationImageToImageMetric, that the histograms are computed in the whole image. 
> 
> For example if I want to register 2 images with a simple translation, and my translation is half the size of the picture, the other half pixels have a default value. Are these pixels taken into account when mesuring the metric?
> 
> Thanks for your answer,
> Francois
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
>