[Insight-users] Registration : Update works but StartRegistration fails

Luis Ibanez luis.ibanez at kitware.com
Tue Aug 30 12:21:27 EDT 2005


Hi Vincent,

Thanks for letting us know that you found the error
in the region setting of your registration code.

--

Regarding your second question
...well, I didn't quite find the question in your email,
only the vague statement that you attempted several
registrations and "...it didn't work...".

As you can understand, we will need a lot more of specific
technical information in order to be able to help you.


1) Did you printed out the trace of metric values and
    parameters, as it is recommended in the ITK Software Guide ?

2) If so, can you post it to the list ?

3) Did you plotted the values of the metric versus
    the number of iterations ?

4) Did you plotted the transforms parameters versus iterations ?

5) Did you saved and viewed the image that your imported from MRI ?
    to verify that the image was imported correctly.

6) Did you saved and viewed the image that you resampled ?
    to verify that the image was resampled correctly

7) Did you initialized the center of rotation of your transform ?

8) You said that you "noised" the parameters of the transform...
    How much noise did you applied ?
    Did you considered the different ranges of values ?
    e.g. angle is measured in radians, while translations are
         measured in millimeters, so for example 0.5 is a huge
         change in the angle, and a minimal one in a translation.


Please give us more technical details,
so we can help you.


Note that you can also post this as a case study to the Insight Journal,
describe your test, post your source code and your input/output images.
What you are doing is a typical test for any new user of the 
registration framework, and having it available as a case study
will help to fine tune your code and to provide guidance to new users.



   Regards,


      Luis



----------------------
Vincent Daanen wrote:
> Hi,
> 
> It's ok right now. Running into Itk source tree in debug mode, I found 
> that registration::Initialize() failed because the region was not set 
> properly.
> 
> Another problem arised then :
>    1/ I'm importing one MRI dataset (a binary file from the disk) with 
> the itk::ImportImageFilter (--> OK)
>    2/ It is then resampled with an arbitrary 3D rigid transform 
> (itk::VersorRigid3DTransform) (--> OK)
>    3/ I noise the rigid transform (axis,angle and translation)
>    4/ I register the two datasets rigidly (SSD, MI, MAttes MI) and I 
> expect to find the initial arbitrary 3D rigid transform but I don't !
> 
> I'm really stick with that because if I'm not able to register a dataset 
> with itself, I'm not sure I will be able to register it with another one !
> 
> Any idea is welcome ..
> 
> Vince
> 
> Luis Ibanez a écrit :
> 
>>
>> Hi Vincent,
>>
>> It is quite likely that you are having an image
>> reading problem.
>>
>> What is the fileformat that you are feeding as
>> input to ImageRegistration8  ?
>>
>> Please give us details about the number of pixels
>> of the Fixed and Moving images in all dimensions,
>> their pixel spacings, and the origin coordinates
>> of the images.
>>
>>
>>   Thanks
>>
>>
>>     Luis
>>
>>
>>
>> ---------------------
>> Vincent Daanen wrote:
>>
>>> Hi all,
>>>
>>> I'm just starting a project where I have to register to MRI datasets.
>>> As I don't want to loose time, I want to use Itk to perform this.
>>> Therefore, I  read ImageRegistration8.cxx and work with this as a basis.
>>>
>>> My problem is the following : when triggering the registration 
>>> process with regitration->StartRegistration(); the program ends with 
>>> the following error
>>>
>>> ExceptionObject caught !
>>>
>>> itk::ExceptionObject (
>>> This application has requested the Runtime to terminate it in an 
>>> unusual way.
>>> Please contact the application's support team for more information.
>>>
>>> If I use regitration->Update, some calculations are performed but I'm 
>>> not sure that it does register the datasets.
>>>
>>> Can anybody points me to clue to solve this error ?
>>>
>>> Thanks
>>>
>>> Vincent Daanen
>>>
>>
> 
> 



More information about the Insight-users mailing list