[Insight-users] Re: ITK ImageRegistration example

[Luis Ibanez]

Hi Chia-Chi

Here are the steps you can follow for running the
ImageRegistration example.

This describes how to use it with the images from the BrainWeb
project available on the ftp site:

   ftp://public.kitware.com/pub/itk/Data/BrainWeb/

In particular:

   Fixed  Image:  brainweb1e1a10f20.mha
   Moving Image:  brainweb1e1a10f20Rot10Tx15.mha

The second image has been rotated intentionally
by 20 degrees around the Z axis and translated 15
pixels along the X axis.  This facilitates testing
registration runs.

Please follow these steps

1) click on the "Load" button beside "Fixed Image" and
    load the file  "brainweb1e1a10f20.mha"

2) when the buttons turn green, click on "Fixed Image",
    a small window will popup with a viewer that allows
    you to visualize the image slice by slice.

3) click on the "Load" button beside "Moving Image" and
    load the file  "brainweb1e1a10f20Rot10Tx15.mha"

4) when the buttons turn green, click on "Moving Image",
    an independet viewer will appear showing this new image.

5) Click on the "Resample Filter" button.  This is intended
    to apply a known rigid transformation to the loaded image.
    Since in this case we are loading an image for which we
    already know the missregistraiton parameters  you can leave
    an identity transform here (this is the default).  If later
    you want to use this feature, you will have to write the
    x,y,z components of a translation on the "Translation" box,
    and the angle "A" and components of the axis vector around
    which you want to apply a rotation. The rotation is applied
    first and then is followed by the translation.

6) When the "Moving Image" button after the "Resample Filter"
    turns green click on it. A viewer with the image will appear.
    It should look just as the image you loaded, since an identity
    transform is being applied by the resample filter.

7) Select the metric you want to use for the registration.
    This is done with the "Metric" choice on top of the big
    "Registration Method" button.  Select "Mean Squares" for
    a first run.

8) Select the Transform for which you want resolve the registration.
    This is done with the "Transform" choice box on top of the
    "Registration Method" button.  Select "Translation" for a first
    run.

9) Select the Interpolator to use when evaluating the moving image.
    This is done with the "Interpolator" choice on the upper right
    corner of the GUI. Please select "Nearest Neighbor" for a first
    run (this will be faster than the linear interpolator).

10) Select the Optimizer that will be used to explore the parameter
     space of the transform and optimize the value of the metric.
    This is done with the "Optimizer" choice box on the right of the
    GUI.  Please select "RegularStepGradientDescent" for a first run.

11) Now that all the component of the registration problem have been
    selected you can start the registration. Click on the "Start" button
    on the lower right of the GUI. The optimization process will start
    at this point. It will take a while, since this example is not
    using the MultiResolution framework.

12) When the registration process is done the "Registration Method"
    button will turn green. At this point you can resample the image
    using the "Resample Filter" on the lower right of the GUI. When
    the filter is done, you could see the result of the registration
    by clicking on the "Registrerd Image" button.




Please let us know if you encounter any problem running the example,

   Thanks

     Luis



======================================================================

Chia-Chi Teng wrote:
> Thanks for the info. I've got the images loaded, but I can't quite figure
> out how to run the application. What are the sequence of steps I need to do
> to run, i.e. what buttons to click after loading the images? If I just click
> start, it crashes. What is the expected output? The readme doesn't really
> quite cover that.
> 
> I'm running the Windows version, by the way.
> 
> Thanks,
> Chia-Chi
> ----- Original Message -----
> From: "Luis Ibanez" <luis.ibanez@kitware.com>
> To: "Chia-Chi Teng" <ccteng@u.washington.edu>
> Cc: "Insight-users" <insight-users@public.kitware.com>
> Sent: Thursday, September 19, 2002 1:44 PM
> Subject: Re: ITK ImageRegistration example
> 
> 
> 
> Hi Chia-Chi,
> 
> The ImageRegistration example uses MetaImages as input.
> 
> This is a simple format using a text header followed by
> the binary data.
> 
> You can download images in this format from the following
> ftp site:
> 
>      ftp://public.kitware.com/pub/itk/Data/BrainWeb
> 
> 
> Note that there are already some pairs of images
> intentionally missregistered. They are intended to
> serve as test cases.
> 
> 
> For example:
> 
> original data in
> 
>      - BrainPart1.tgz
> 
> is rotated 10 degrees and translated 15 pixels in:
> 
>      - BrainPart1Rotated10Translated15
> 
> and rotated 20 degrees and translated 20 pixels in
> 
>      - BrainPart1Rotated20Translated20
> 
> 
> You can also generate your own images using the MetaImageImporter
> utility that is build as part of the ITK examples. You will
> find it on the directory:
> 
>       Insight/Examples/MetaImageImporter
> 
> 
> Please let us know if you encounter any problems,
> 
> 
> Thanks
> 
> 
> 
>    Luis
> 
> 
> =============================================================
> 
> Chia-Chi Teng wrote:
> 
> 
>>Hi, I'm trying to use the ITK to do some medical image registration. I saw
>>this ImageRegistration example and want to try it, but couldn't find any
>>sample images, and couldn't figure out what format of image it takes as
>>input. I saw your name in the readme file, hopeful you can help me with
>>
> the
> 
>>questions. Or do you have any sample images that I can use to try this
>>example app?
>>
>>Thanks,
>>Chia-Chi
>>
>>
>>
>>
> 
> 
> 
> 
>