[Insight-users] problem with GDCM reader and linux

[Mathieu Malaterre]

C. Aaron Cois wrote:
> I'm using GDCMImageIO to read some dicom series from data collected on a 
> GE machine.  I'm having no trouble with this on my Windows machine.  
> However, when I try to read in the same files, using the exact same 
> software compiled on a Suse Linux machine (which I need to do, as this 
> is our heavy computation machine), the images are read in, but the data 
> is incorrect.  All pixels are read in as the same intensity, though the 
> size, spacing and origin are all read in correctly.
> I have tested to see whether this was an issue of file corruption from 
> transfer from the windows machine to the linux machine, and have 
> confirmed that it is not (by transferring the files back from the linux 
> machine to the windows machine and reading them in without issue).  The 
> only thing I can think of is that there is some issue with GDCMImageIO 
> being compiled on my linux platform making it function improperly.  
> Anyone have any ideas?


Aaron,

	I am very surprised of the result as my main platform of development 
for GDCM is indeed Linux. I never played with a Suse though.
	Depending on how much time you have on your hand here is a couple of 
suggestion:

* First thing first, does `ctest -R GDCM -V` works for you ? (run all 
the GDCM related tests on the Suse machine).
* Strip down the test to the minimal test, like ImageReadWrite and run 
it through valgrind.
* Download GDCM from it's main CVS server (*). And build the examples. 
PrintFile will dump the header and if you build GDCM with USE_VTK:ON 
you'll get a vtkgdcmViewer which should display the DICOM image file.

If all else fails, send me the original DICOM image, the DICOM image 
rewritten by GDCM on the Suse and the C++ code used to rewrite it. I'll 
inspect what can be wrong.

BTW is this a 64 bits system ? What is the ITK version used (on windows 
and Suse) ?
Mathieu
(*)
http://www.creatis.insa-lyon.fr/Public/Gdcm/Cvs.html