[Insight-users] Re: registration: CT to PET

[Luis Ibanez]

Hi Soumed,

The PNG that you sent has a saturation in the
gray level values. You may have to rescale
the intensities in order to fit in 16 bits.

Please read the chapter on ImageRegistration
from the SoftwareGuide and once you feel
confortable with the concepts start your
registration by using the example:


      ImageRegistration4.cxx


that deals with Multi-modality images.


Note that the two slices that you have do not
necesarily will register since they not necessarily
belong to the same plane in the actual anatomy.

You should perform this registration in 3D using
the full datasets in order to account for rotations
and translation in 3D space.

Whenever you need to verify the quality of the
resulting transform parameters, you simply use
them in the ResampleImageFilter and ovelay the
fixed image on top of the resampled moving image.

You will find examples on this in almost all the
.cxx files in Insight/Examples/Registration/




   Regards,



      Luis



---------------------
S Dey wrote:

> Hi!!!!
> Here I am sending the Image-110_0.png file(converted from Image.-110_0.dcm)
> .
> Plz teel the correct registration result.
> Waiting for ur reply
> thnx.
> soumen
> 
> 
> -----Original Message-----
> From: insight-users-admin@itk.org [mailto:insight-users-admin@itk.org]On
> Behalf Of Luis Ibanez
> Sent: Sunday, April 25, 2004 11:49 PM
> To: soumend@magnum.barc.ernet.in
> Cc: Insight-users@itk.org
> Subject: [Insight-users] Re: registration: CT to PET
> 
> 
> Hi Soumend
> 
> Your CT DICOM image 1188316 was ok,
> but the Image.-110_0.dcm was not
> readable by ITK readers, nor by
> MRIConvert.
> 
> Could you please resend it in another format ?
> 
> The scheme that you are using doesn't seem
> to be reasonable. Note that you can do this
> registration in 3D, no need to do it slice by
> slice. It is almost imposible that you will be
> able to follow the 3D trajectories of the actual
> images in 3D just by manually  tracking the
> correspondences between slices in one image and
> slices in the other image.
> 
> To put it short:   Please do this in 3D.
> 
> You will find multiple examples of Multimodality
> registration in the SoftwareGuide
> 
>    http://www.itk.org/ItkSoftwareGuide.pdf
> 
> and the associated source code is available on
> the Insight/Examples/Registration directory.
> 
> ---
> 
> In order to verify if the final transform parameters
> are appropriate, you simply use those parameters for
> setting up a transform and use it in the ResampleImage
> Filter in order to reformat the Moving image into the
> space of the FixedImage.
> 
> Almost all the examples in Insight/Examples/Registration
> do this resampling at the end.
> 
> Once you have resampled the moving image, you use any
> visualization tool in order to superimpose the fixed
> image and the resampled moving image.  You could do
> this directly by using VolView. The free version of
> VolView can be downloaded from:
> 
>    http://www.kitware.com/products/volview.html
> 
> Or, if you prefer to write your own code, you could
> use VTK classes such as the vtkImageBlend filter and
> the vtkImageViewer.
> 
> 
>     Regards,
> 
> 
>        Luis
> 
> 
> -----------------------------------
> soumend@magnum.barc.ernet.in wrote:
> 
>>hi!!!!
>>
>>I have wriiten a program to register one CT image(512X512)with one PET
> 
> Image
> 
>>(128X128). But the problem is I donno whether the transform parameters I
> 
> am
> 
>>getting as output after registration are right or not. So I am sending the
> 
> two
> 
>>image files as attachments plz let me know the transform parameters which
>>result correct registration. I am using Mattes MI metric for registration.
>>
>>Anothe thing ....I am following the scheme stated below...
>>
>>Actually I have 116 CT images and 163 Pet images. I am taking fixed slice
> 
> name
> 
>>from the user and finding the moving slice from the other directory.To do
> 
> that
> 
>>I am using the relative position of fixed slice from the furthest slice in
> 
> its
> 
>>series as guide. The moving slice which matches best with fixed slice in
> 
> its
> 
>>relative position wrt furthest slice in its series gets seleceted for
>>registration.
>>eg.
>>fixed slice 1188316 (CT iamge, Slice Location=544, furthest slice in
>>series=654, relative posn=110)
>>moving slice gets selected Image.110_0.dcm (PET image,slice location=-
>>110.5,furthest slice in series=0, relative posn=110.5)
>>
>>Is this scheme reasonable? It works fine for my present data. But I want
> 
> to
> 
>>know what are the drawbacks of this scheme.
>>
>>Ur Reply is eagerly awaited.
>>Thnx.
>>
>>Soumen
>>
>>
>>-------------------------------------------------
> 
> 
> 
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users@itk.org
> http://www.itk.org/mailman/listinfo/insight-users
> 
> 
> ------------------------------------------------------------------------
>