[Dcmlib] gdcm : Re: AW: Contact [195.220.108.240]

[Jean-Pierre Roux]

Bonsoir.

En cherchant des images 'true Dicom Big Endian' pour pouvoir tester une 
partie dans laquelle nous ne sommes jamais passés faute de données, 
j'etais tombé sur un site (julius.caesar.de) qui contenait qq series 
d'images DICOM ... qui petaient toutes avec gdcm (ainsi qu'avec les 
autres viewers).
Classique, mais inquietant -as usual- !

Ce qui l'est moins (classique), c'est qu'ils ont *fait expres* de mettre 
des bugs dans l'entete, pour eviter qu'il y ai trop de gens qui 
downloadent leurs données.
(voir reponse ci apres)
Amusant, non ?

JP


Thomas Jansen wrote:

>Hi Jean-Pierre,
>
>: We downloaded images from your Data Set.
>: They break all the Dicom Viewers we have (e-film, xmdecon 0.8.13, ...)
>: According what we can read with our own DicomReader (gdcm), the images
>were
>: written by xmedcon 0.8.8, so we are supprised that xmdecon 0.8.13 breaks.
>:
>: Moreover,  just after the Dicom Element (0054,0200), there is a DICOM
>: element (0000,0000) whose VR is SQ; this is the point that seems to cause
>: trouble with DICOM viewers : element 0000 of each group is (was?) supposed
>: to contain group length ...
>:
>: Would you have any information about that ?
>
>You're right! The datasets are not in 100% valid DICOM format.
>
>The datasets should only be used in the Julius environment: The "special"
>DICOM format keeps away massive download traffic from our server from those
>who are only interested in the datasets.
>
>However, all datasets from the “Chapel Hill Volume Rendering Test” (Volume I
>& II) can be downloaded from
>http://www.unf.edu/public/cap6400/ychua/3dslice/ (in raw format). Use the
>free XMedCon (http://xmedcon.sourceforge.net/) tool to convert them to
>DICOM.
>
>Regards,
>--Thomas
>
>
>  
>