JPEG XR - DICOM

This is a discussion on JPEG XR - DICOM ; Does anyone know if there are any plans to support JPEG XR in the Dicom spec? Thanks Jim...

+ Reply to Thread
Results 1 to 6 of 6

Thread: JPEG XR

  1. JPEG XR

    Does anyone know if there are any plans to support JPEG XR in the
    Dicom spec?

    Thanks

    Jim


  2. Re: JPEG XR

    On Aug 8, 11:29 am, jim wrote:
    > Does anyone know if there are any plans to support JPEG XR in the
    > Dicom spec?


    Is it even a standard yet ? I cannot find any publication about JPEG
    XR in pubmed, how well does it perform for medical images ?

    -Mathieu


  3. Re: JPEG XR

    Hi,

    MS have unoffically said it's about to be an ISO.

    I know MS are pushing it quite hard, and am currently looking for
    answers to:

    - are the dicom jpeg formats supported in JPEG XR?
    - if so, is anyone currently talking about it for use in dicom?

    Thanks

    Jim


  4. Re: JPEG XR

    > Does anyone know if there are any plans to support JPEG XR in the
    > Dicom spec?


    I don't think so, although I have to admit that I have not followed the progress of DICOM WG5,
    which would be responsible for such a DICOM extension (definition of a new transfer syntax)
    recently.

    As a matter of policy, DICOM will only ever adopt encoding schemes that are standardized,
    and for JPEG-XR there are only intentions of standardisation, work has to my knowledge not
    even started yet - so far we are speaking of a proprietary compression scheme named "HD Photo".

    That said, even if JPEG-XR should be standardised, the question would still be whether
    - JPEG-XR provides a significant benefit over the existing compression schemes
    (JPEG, JPEG 2000, JPEG 2000 Part 2, JPEG LS) that are already adopted in DICOM
    when used for compression of medical imagery
    - there is a significant interest from industry to add JPEG-XR support to their products.

    Uptake of JPEG 2000 in the DICOM world has been rather slow, arguably because the advantage
    of lossy JPEG 2000 compression over traditional DCT based JPEG is not as large as one might
    think in the quality range usually employed with medical imagery. JPEG 2000 is certainly
    much better than JPEG at extremely low bitrates, but these are not used in the medical world
    anyway, and an improvement of the compression factor from, say, 5:1 to 8:1 is not worth the hassle.
    For lossless compression, JPEG-LS already provides a significant improvement over JPEG lossless
    and requires much less computational complexity than JPEG 2000 lossless, nevertheless uptake
    has been even slower than JPEG 2000 in the DICOM world, as far as I can see.

    Regards,
    Marco Eichelberg
    OFFIS

  5. Re: JPEG XR

    On Aug 9, 9:29 am, Marco Eichelberg
    wrote:
    [snip]
    > As a matter of policy, DICOM will only ever adopt encoding schemes that are standardized,
    > and for JPEG-XR there are only intentions of standardisation, work has to my knowledge not
    > even started yet - so far we are speaking of a proprietary compression scheme named "HD Photo".


    There will be a ballot early October on whether to go ahead with
    standardisation. Then (if the vote is "yes") standardisation may take
    about a year.
    http://www.barrypearson.co.uk/articles/hdp/history.htm

    > That said, even if JPEG-XR should be standardised, the question would still be whether
    > - JPEG-XR provides a significant benefit over the existing compression schemes
    > (JPEG, JPEG 2000, JPEG 2000 Part 2, JPEG LS) that are already adopted in DICOM
    > when used for compression of medical imagery

    [snip]

    My own tests show there is a significant improvement over JPEG
    compression - perhaps up to a factor of 2 in size, and fewer
    artifacts:
    http://www.barrypearson.co.uk/articl...ysis_lossy.htm

    Size after compression appears similar to JPEG2000. Microsoft claim
    that compression is faster than JPEG2000. (I wasn't able to measure
    this).

    I did a few tests of lossless compression here:
    http://www.barrypearson.co.uk/articles/hdp/analysis.htm

    --
    Barry Pearson
    http://www.barrypearson.co.uk/photography/


  6. Re: JPEG XR

    Barry Pearson wrote:
    > My own tests show there is a significant improvement over JPEG
    > compression - perhaps up to a factor of 2 in size, and fewer
    > artifacts:
    > http://www.barrypearson.co.uk/articl...ysis_lossy.htm


    This is about the same advantage that is usually claimed for JPEG 2000
    over JPEG. However, I noted that your tests have been performed with
    digital photographies and you have used "traditional" measures such
    as histogram deviation to determine image quality. None of that necessarily
    translates to medical imaging. The physical properties of CR, CT or MR
    images are quite different from photographies and the compression results
    may or may not be in the same order of magnitude. Also measures such
    as histogram comparison or SNR are very poor indications of image quality
    from a medical perspective - it is not relevant whether or not an image "looks good",
    it is relevant whether or not the compression possibly masks relevant
    information. Metrics based on JNDs (just noticeable differences) have been
    found to be much better for this purpose.

    > Size after compression appears similar to JPEG2000. Microsoft claim
    > that compression is faster than JPEG2000. (I wasn't able to measure
    > this).


    For lossless compression, the main competitor will be JPEG-LS, not JPEG 2000 -
    it would be interesting to see how JPEG XR compares here, because JPEG-LS
    was explicitly designed for low complexity (=fast) lossless compression.

    I personally would be rather surprised if the medical world would happily
    jump on the JPEG XR bandwagon - it may be somewhat better than JPEG 2000
    if you look at the combination of complexity and compression factor, but
    I doubt that the advance over JPEG 2000 will be worth the effort - JPEG XR would
    need to provide an improvement of a factor of 2 over JPEG 2000 to be considered
    a real improvement. But that's just my personal view.

    Regards,
    Marco Eichelberg

+ Reply to Thread