dcm2xml and xml2dcm

[retsil]

I've been using xml2dcm and dcm2xml to prototype a simple DICOM reader/writer/editor.

Most of my work invloves using an existing DICOM file as a template. Thus, I've reasoned that I can probably make of the modifications I would ever need to using XPaths.

I was interested if there is a convention for defining an XPath such that it always refers to a unique element.

I've considered that the 2-byte tag is unique and can only be repeated in different groups or sequences


Excerpt from dcm2xml.dtd

Code: Select all

<!ATTLIST element tag CDATA #REQUIRED > # This is a unique key

<!ATTLIST sequence tag CDATA #REQUIRED > # This is a unique key

An example XPath might be

Code: Select all

/child::file-format/child::data-set/child::sequence[attribute::tag="0054,0410"]/child::item[position()=1]/child::sequence[attribute::tag="0054,0412"]/child::item[position()=1]/child::element[attribute::tag="0008,0100"]

Would anyone be able to enlighten me that I am (reasonably) correct in these assumptions?

[Jörg Riesmeier]

I've considered that the 2-byte tag is unique and can only be repeated in different groups or sequences

The tag (group and element number) is unique within an item/dataset. This is at least true for valid DICOM objects.

[retsil]

Just a further question Jörg,

Is base64 encoding/decoding supported for all transfer syntaxes or does it just work for some?
I can't get dcm2xml to generate base64 encoded tags (where VR is OW or OB) when using the +Eb (aka --encode-base64) on SunOS

Cheers,

Robbie

[Jörg Riesmeier]

You also need to specify +Wb (--write-binary-data). In case you want to encode all binary data (even the pixel data) use +M (--load-all) in addition -- but be careful with large images!