[Insight-developers] Current limitation of ITK for writting DICOMfiles

[Lorensen, William E (Research)]

We know about this problem. jim Miller and I are working on it. For now, if you want to write a correct dicom series you need to do it one file at a time and populate the image position patient field.

I'll talk to Dirk about this Monday. I've been out of th country for a week.

Bill

-----Original Message-----
From: insight-developers-bounces+lorensen=crd.ge.com at itk.org
[mailto:insight-developers-bounces+lorensen=crd.ge.com at itk.org]On Behalf
Of Mathieu Malaterre
Sent: Friday, May 20, 2005 11:59 AM
To: Insight Developers
Cc: Padfield, Dirk R (Research)
Subject: [Insight-developers] Current limitation of ITK for writting
DICOMfiles


Hello,

	I was recently contacted by Dirk Padfield (GE Research), he had 
troubles writting a series of DICOM files. In a very simple code he 
demonstrates that eventhough he has access to the input DICOM headers it 
is completely useless for writting the output serie. In his example he 
makes use of the itk::OrientImageFilter filter. So the input is let say 
256x256x3 and the output is 256x3x256. Therefore he has to generate 256 
DICOM headers from only 3 in input.

	I realize that currently ITK SeriesWriter instanciate a 2D writer in 
this case and therefore information about the 3d dimension is completely 
lost: preventing us to generate any appropriate 
orientation/spacing/positions...

	Is there any work done for dealing with this problem ? How does other 
people deal with this issue ?


	The only solution I came with was to hand-generate an array of 
MetaDataDictionary containing the Image Number. Ideally it should also 
contains Image Patient Position / Image Patient Orientation . See 
attached file where I patched Dirk's code.


Thanks
Mathieu