[Dcmlib] [Fwd: RE: GE Implicit VR Big Endian]

[Mathieu Malaterre]

For information: GDCM is not really ready for GE Implicit VR Big Endian, 
luckily so far we only had 8bits images -guess why it was working :)-

Could anyone provide such image ?
Mathieu

-------- Original Message --------
Subject: RE: GE Implicit VR Big Endian
Date: Thu, 6 Oct 2005 10:13:49 -0400
From: Miller, James V (Research) <>
To: Mathieu Malaterre <>

Mathieu,

(Note that we just checked in an additional fix a few minutes ago).  I 
was going to
send you a note on this once the dashboard cleared up.  We probably need 
to have
similar changes put into writing the data.  I added a long comment in 
this code
on the particulars of this format.  Basically, header is little endian and
pixel data is big endian.

I only discovered this problem yesterday when someone switched their 
application from
DICOMImageIO2 to GDCMImageIO.  They needed to use GDCMImageIO so that 
they get the
direction cosines but GDCM was not handling the byteswapping appropriately.

I am not sure that I have an image that I can redistribute.  I'll dig 
around a bit.

Jim

-----Original Message-----
From: Mathieu Malaterre [mailto:]
Sent: Thursday, October 06, 2005 10:03 AM
To: Miller, James V (Research)
Subject: GE Implicit VR Big Endian


Jim,

	By any chance do you have a copy of this image. The only one we have
was written on 8bits... and thus the bug was never seen. I would like to
add it to the test suite.

Thanks,
Mathieu