[Insight-users] Intensity values issue

[Michael Pozdnyakov]

Hi Luis,

First of all, thank you for your answer.

Actually my problem had been already solved, I only haven't written  
yet about it.

To debug this issue, first of all I tried, as you said, to write  
directly the image I read. Result, no more problem. So then I tried to  
read pixels of output images on each stage of my code. And I  
discovered, that intensity values never become negative. So, the  
problem was in writing. After some research, I noticed, that there was  
RescaleSlope tag value of Dicom Meta Data of the fixed image, who was  
smaller than the moving image's one. And after I changed it (I took  
the value of moving image) I had no more problem.

But it is always strange for me, that there is this kind of casting.  
Is there any explanation?

Thanks,
Mikhail
On 25 May 2010, at 17:29, Luis Ibanez wrote:

> Hi Michael,
>
> Thanks for the detailed description of your problem.
>
> Looking at your code, it seems that you are using a "float" pixel
> type, not a "double" pixel type, as you describe in your email.
>
>
> It is unlikely that this change in pixels values is produced by
> the resampling process itself.
>
>
> Could you please do the following:
>
> 1) Connect the Series reader output directly as input to the
>    Series writer (that is, skip the resampling process altogether)
>
> 2) Look at the resulting output image, and let us know if that
>    image does have the pixel values changed too.
>
>
>   Thanks
>
>
>      Luis
>
>
> -----------------------------------------------------------------------------------
> On Mon, May 17, 2010 at 9:16 AM, Michael Pozdnyakov <mishgunn at gmail.com 
> > wrote:
>> Hello,
>>
>> I'm working now on 3D rigid registration task and I have some  
>> issues with my
>> program at the moment of transformation. For the better explication  
>> of the
>> issue I will describe only transformation part.
>>
>> I have a body PET exam in a series of DICOM files. I read this  
>> series with
>> GDCM and the reading goes well and I keep a 3D image in memory (in  
>> double).
>> Then I create a ResampleImageFilter to transform the image. I  
>> define the
>> transformation (I've tried many of them: translation, versor rigid,  
>> affine,
>> euler and even identity) and I define the interpolator. And finally  
>> I save
>> my transformed image in another DICOM series. Everywhere I have  
>> double data
>> type.
>>
>> So, my issue is that high intensity values (bigger than ~16K) become
>> negative. And it doesn't depend on transformation type neither on
>> interpolator. I have always the same probleme in high intensity  
>> zones.
>>
>> Thanks,
>> Mikhail
>>
>> PS.
>> My configuration
>> MacOS X 10.5.8, gcc 4.0.1
>> ITK 3.16
>>
>> PPS
>> Sources in attachement
>>
>>
>>
>> _____________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at
>> http://www.kitware.com/opensource/opensource.html
>>
>> Kitware offers ITK Training Courses, for more information visit:
>> http://www.kitware.com/products/protraining.html
>>
>> Please keep messages on-topic and check the ITK FAQ at:
>> http://www.itk.org/Wiki/ITK_FAQ
>>
>> Follow this link to subscribe/unsubscribe:
>> http://www.itk.org/mailman/listinfo/insight-users
>>
>>