[Dcmlib] [Fwd: Re: [Insight-developers] DICOM Header Tags]

[Mathieu Malaterre]

Just for google cache:

About the famous SIEMENS (0029 1020) DICOM tag. One of the few private 
dicom tag actually needed to reconstruct the image for further analysis.
To make things worse this is a *very* long binary field which is not 
loaded by default due to the default 0xfff length restriction in gdcm.



-------- Original Message --------
Subject: Re: [Insight-developers] DICOM Header Tags
Date: Tue, 16 Aug 2005 11:00:10 -0400
From: Mathieu Malaterre <>
To: vincent-magnotta
References: <> <1124202873.14974.12.camel at tesla.radiology.uiowa.edu>

Thanks a bunch for the link ! I'll keep that

Vincent A. Magnotta wrote:
> Mathieu,
> 
> Thanks for your help. I will give your suggestion a try.This field as
> actually an ASCII parameter map for the Siemens pulse sequence used to
> create the images. It provides all information allowing you to take an
> image from one Siemens scanner to another scanner and completely specify
> the imaging parameters. An example of some of the fields are shown at
> this site:
> 
> http://www.enac.northwestern.edu/~tew/archives/2003/02/25/incomplete-dicom-headers/
> 
> 
> Since information about the DTI study such as number of directions and
> number of b values applied are stored in this header, it will allow me
> to convert the 3D data to a 4D representation that we use to analyze the
> data.
> 
> Thanks Again,
> 
> Vince
> 
> 
> On Tue, 2005-08-16 at 08:52, Mathieu Malaterre wrote:
> 
>>Ah ok I see your problem :)
>>
>>sorry there is 'feature' in gdcm where long binary entry are skipped to 
>>decrease memory usage:
>>
>>$ grep MAX_SIZE_LOAD_ELEMENT_VALUE gdcm/*
>>
>>It is currently set to 0xfff, whereas in your case you have an entry 
>>0xc940 long (51520). The easy work around is then change this line in gdcm:
>>
>>gdcmDocument.cxx:const unsigned int 
>>Document::MAX_SIZE_LOAD_ELEMENT_VALUE = 0xfff; // 4096
>>
>>to something bigger.
>>
>>Or at the itkGDCM level force loading this value.
>>
>>But I definitely agree this is not easy to do. I'll assign me a bug to 
>>provide the API from itkGDCM to set this value.
>>
>>Sorry for trouble
>>Mathieu
>>Ps: This entry is really long, what does it contains? In fact this is 
>>the first time a binary shadow entry will be used and you seems to be 
>>able to decode it.