Group Length and Private Group

[Mathieu Malaterre]

Hi,

According to DICOM Standard (Part 5, Section 7.2), Group Length are of
the form (gggg,0000). But I recently came across this image:

http://nova.nlm.nih.gov/WRAMC/DICOM/WRAMC%20VC-443/225017318462C331%20F2EA3F46FAE7321B/99B7F85B3D1343698B6D301746477DEF/2F612AF2451EEDF03AC73C7D19AF8033/FFBE146B651B2A44C6ADC12DB614E3ED.dcm

Where I get:

V 130d|0000 [UL]
[Group Length] [1099164365\1276292990\3804772504\555532589] x(4183eacd)
V 140d|0000 [UL]
[Group Length] [835531882\1219434392\151033222\3313626383] x(31cd346a)
V 150d|0000 [UL]
[Group Length] [4259307444\1128414896\4113509537\2144277863] x(7fffffff)
V 160d|0000 [UL]
[Group Length] [3824301659\1114577058\3573223579\2679727550] x(7fffffff)
V 170d|0000 [UL]
[Group Length] [3850819895\1116855751\1080659125\3543366330] x(7fffffff)

These clearly does not define Group Length (they are not even UL). Has
anyone seen this before ? How am I suppose to read/interpret those ?

Thanks
Mathieu

[Peter B Schmidt]

Mathieu,

this looks strange - like someone tried to enter private data without
knowing anything about DICOM private group handling.

I am pretty sure that this is not directly taken from the GE scanner,
some software must have messed this up.

- No private group creator
- VM != 1 in Group length attributes (which are retired, afaik)

hm.

I would be curious to learn about the tale of woe this image has gone
through, do you have any details about that?

Regards,

Peter

[Mathieu Malaterre]

Peter,

Thanks for taking the time to answer.
No I was just browsing D. Clunie website in quest for some potential
DICOM image that would not be readable by gdcm. And just found out this
website, and took one image...

I guess you are right there might have been some post-processing that
skrewed things up.

Mathieu
Ps: If you want to see a real brain dead 'DICOM' image that a user sent me:
http://www.creatis.insa-lyon.fr/~malaterre/gdcm/1075235.dcm
Everything is just so wrong :)