[Dcmlib] Philips images

[Mathieu Malaterre]

JP,

	That's interesting !
	Anyway I don't believe right now we need to read them, what I would 
like to see is that gdcm peacefully return saying something like:

-> Sorry the lenght of the sequence is invalid, dont't know how to proceed.

	If we are lucky enough by reading the length of the sequence we can 
detect this is outside of the file... otherwise we need to detect that 
the value is somehow absurd.

	Is there any example available I would like to raise the issue on 
news.protocol.dicom, to see how other people deal with it.
	As a side note, I'd like to try TomoVision on this image.

Mathieu
Ps: JP can you fix your warning:

  warning C4305: 'initializing' : truncation from 'const int' to 'short'
http://public.kitware.com/GDCM/Testing/Sites/Creatis.vc60/GDCM-Win32-vs60-Dbg/20050629-0100-Nightly/BuildWarning.html

Jean-Pierre Roux wrote:
> Guest what?
> I've got new gdcm-breaker Philips images.
> As usual (?) they break also the other viewers :-(
> Still the bloody stuff around endianess switching .
> 
> Now, it's *really* unpredictable !
> We used the delimiters (dicom group 0xfffe) to guess wether the 
> endianess was actually the one specified in the meta elements, and 
> switch the 'swap code' when necessary.
> And it worked !
> 
> Within these new images (FLAIR sequence), the first delimiter inside a 
> 'shadow' sequence show us the endianess just switched (but the lengthes 
> don't)
> --> Segfault !
> I can't imagine an heuristics that would deal with both cases ...
> 
> The only way to read theses images is to skip private sequences.
> (we can use a new value for LoadMode)
> 
> I think people writting clinical interfaces (Benoit, Emmanuel, ...) 
> should allow end user to tell whether he wants to skip any Sequence 
> (LoadMode = NO_SEQ), to skip any shadow group (LoadMode = NO_SHADOW), or 
> to skip shadow Sequences LoadMode = NO_SHADOWSEQ).
> 
> JP
> 
> 
> _______________________________________________
> Dcmlib mailing list
> Dcmlib at creatis.insa-lyon.fr
> http://www.creatis.insa-lyon.fr/mailman/listinfo/dcmlib
>