DICOM wrapping and unwrapping

[Karl H]

Hi All,

   I'm working with a group that is collecting multi-modality data - imaging and non-imaging. The imaging data are all in DICOM format, the non-imaging (e.g., EEG) is not. There is an ongoing discussion about whether to DICOM wrap the non-DICOM data and store it in the same store as the native DICOM data (so the data is all in one place, so we can query one location to build cohorts, etc). We have a tool to DICOM wrap the data, but how does one unwarp the data? Figure out how big the DICOM header is for each file and remove that? I haven't found a software package for that online.

Any information welcome,

Karl

[Victor Derks]

Hi Karl,

There are many things to consider to store non-imaging data in DICOM or not. I can provide some benefits, but a lot depends on the actual context of your use case.

If the non-imaging data fits in the DICOM (medical) data model it is very practical to wrap/convert it to the DICOM format:

- Having only 1 data format is easier to maintain then multiple formats.
- It makes it possible to leverage the functionality provided by general purpose DICOM equipment: storage, query, transfer, view
- DICOM has broad support for non-imaging "objects": PDFs, Structured reports, ECG waveforms, video, etc. All very well documented 

- If an existing "object type" is not a good fit for your non-imaging data, DICOM has well defined ways to define private extensions.
- There exists a wide range of commercial and open source software to process DICOM files in different programming languages.

[Karl H]

Hi Victor,

   Thanks that's helpful. However, even if DICOM has support for non-imaging objects, the ultimate data consumers are not used to getting their data in DICOM format and their analysis tools of choice don't handle DICOM. Can anyone point me to code or tools that unwrap the data?

thanks,

Karl

[Brian O'Brien]

Have a look at the RAW IOD.... you might add the file type somewhere in the meta data.

dcmtk and dcm4che have tools for you.

[David Clunie]

There is a standard IOD for EEG, so you should not use the Raw Data IOD.

Since it is one of the (many) waveform IODs that use Waveform Data within a Waveform Sequence item to encode the bulk data, obviously specific tools are required for conversion to and from DICOM to other formats (be they proprietary or academic).

I.e., this is not simply a "wrapping" question, at least not if you want the result to be standard and interoperable.

See, for example, this recent paper:

DICOM® integrated EEG data: A first clinical implementation of the new DICOM standard for neurophysiology data

David