Enhanced MR - DICOM

This is a discussion on Enhanced MR - DICOM ; Please tell me how create and Enahnced MR image form two existing MR images....

+ Reply to Thread
Results 1 to 7 of 7

Thread: Enhanced MR

  1. Enhanced MR

    Please tell me how create and Enahnced MR image form two existing MR
    images.


  2. Re: Enhanced MR

    On 1 Mrz., 08:54, "Shafiq" wrote:
    > Please tell me how create and Enahnced MR image form two existing MR
    > images.


    Shafiq, this is not a very easy task. Just as an outline you would
    need to:
    1. Merge the pixel data to your new Enh object. This will be a
    multiframe format (see Part 5).
    2. Copy identical attributes like patient name and so on to the Enh.
    object
    3. Find attributes belonging to the multi frame functional groups and
    3a. Copy all attributes which are identical in both source objecs to
    the share functional groups
    3b. Copy all attributes different in both source objects to the per
    frame functional groups
    6. Add all required tags to your Enh. object still missing /empty
    depending on type 1 /2

    The real problem is that some of the attributes have different names /
    tag numbers so you need kind of a translation table. So what I refer
    to as "copy all attributes" is actually a complex mapping.
    I think David has a tool which does a basic transformation.

    Thomas



  3. Re: Enhanced MR

    Thomas,

    Thanks for your guidlines. I am newbie to DICOM and have lots of basic
    doubts. Some of the listed are below.

    1. Is it necessary that all the MR images in a Enhanced MR should fall
    under same series.

    2. Is it necessary that all the MR images in a Enhanced MR should of
    Same Dimension, For. e.g. Rows, Columns, Samples Per Pixel etc should
    have same values or not.

    3. What is the name of tool DAvid have, will it run on Windows and
    where can I get it?





    On Mar 2, 8:29 pm, "Thomas Freier" wrote:
    > On 1 Mrz., 08:54, "Shafiq" wrote:
    >
    > > Please tell me how create and Enahnced MR image form two existing MR
    > > images.

    >
    > Shafiq, this is not a very easy task. Just as an outline you would
    > need to:
    > 1. Merge the pixel data to your new Enh object. This will be a
    > multiframe format (see Part 5).
    > 2. Copy identical attributes like patient name and so on to the Enh.
    > object
    > 3. Find attributes belonging to the multi frame functional groups and
    > 3a. Copy all attributes which are identical in both source objecs to
    > the share functional groups
    > 3b. Copy all attributes different in both source objects to the per
    > frame functional groups
    > 6. Add all required tags to your Enh. object still missing /empty
    > depending on type 1 /2
    >
    > The real problem is that some of the attributes have different names /
    > tag numbers so you need kind of a translation table. So what I refer
    > to as "copy all attributes" is actually a complex mapping.
    > I think David has a tool which does a basic transformation.
    >
    > Thomas




  4. Re: Enhanced MR

    Hi

    No, yes, dcmulti in dicom3tools (though it is incomplete), no (except
    under cygwin possibly), my web site.

    David

    Shafiq wrote:
    > Thomas,
    >
    > Thanks for your guidlines. I am newbie to DICOM and have lots of basic
    > doubts. Some of the listed are below.
    >
    > 1. Is it necessary that all the MR images in a Enhanced MR should fall
    > under same series.
    >
    > 2. Is it necessary that all the MR images in a Enhanced MR should of
    > Same Dimension, For. e.g. Rows, Columns, Samples Per Pixel etc should
    > have same values or not.
    >
    > 3. What is the name of tool DAvid have, will it run on Windows and
    > where can I get it?
    >
    >
    >
    >
    >
    > On Mar 2, 8:29 pm, "Thomas Freier" wrote:
    >> On 1 Mrz., 08:54, "Shafiq" wrote:
    >>
    >>> Please tell me how create and Enahnced MR image form two existing MR
    >>> images.

    >> Shafiq, this is not a very easy task. Just as an outline you would
    >> need to:
    >> 1. Merge the pixel data to your new Enh object. This will be a
    >> multiframe format (see Part 5).
    >> 2. Copy identical attributes like patient name and so on to the Enh.
    >> object
    >> 3. Find attributes belonging to the multi frame functional groups and
    >> 3a. Copy all attributes which are identical in both source objecs to
    >> the share functional groups
    >> 3b. Copy all attributes different in both source objects to the per
    >> frame functional groups
    >> 6. Add all required tags to your Enh. object still missing /empty
    >> depending on type 1 /2
    >>
    >> The real problem is that some of the attributes have different names /
    >> tag numbers so you need kind of a translation table. So what I refer
    >> to as "copy all attributes" is actually a complex mapping.
    >> I think David has a tool which does a basic transformation.
    >>
    >> Thomas

    >
    >


  5. Re: Enhanced MR

    David,

    Thanks for answering my doubts.

    1. I downloaded dicm3tools, could not find a windows version of
    dcmulti.

    2. Can I append the pixel data of each slice one by one in final
    target Enhanced MR. Or should i need to do any process. Looking for
    some help on this.





    On Mar 6, 4:42 pm, David Clunie wrote:
    > Hi
    >
    > No, yes, dcmulti in dicom3tools (though it is incomplete), no (except
    > under cygwin possibly), my web site.
    >
    > David
    >
    >
    >
    > Shafiq wrote:
    > > Thomas,

    >
    > > Thanks for your guidlines. I am newbie to DICOM and have lots of basic
    > > doubts. Some of the listed are below.

    >
    > > 1. Is it necessary that all the MR images in a Enhanced MR should fall
    > > under same series.

    >
    > > 2. Is it necessary that all the MR images in a Enhanced MR should of
    > > Same Dimension, For. e.g. Rows, Columns, Samples Per Pixel etc should
    > > have same values or not.

    >
    > > 3. What is the name of tool DAvid have, will it run on Windows and
    > > where can I get it?

    >
    > > On Mar 2, 8:29 pm, "Thomas Freier" wrote:
    > >> On 1 Mrz., 08:54, "Shafiq" wrote:

    >
    > >>> Please tell me how create and Enahnced MR image form two existing MR
    > >>> images.
    > >> Shafiq, this is not a very easy task. Just as an outline you would
    > >> need to:
    > >> 1. Merge the pixel data to your new Enh object. This will be a
    > >> multiframe format (see Part 5).
    > >> 2. Copy identical attributes like patient name and so on to the Enh.
    > >> object
    > >> 3. Find attributes belonging to the multi frame functional groups and
    > >> 3a. Copy all attributes which are identical in both source objecs to
    > >> the share functional groups
    > >> 3b. Copy all attributes different in both source objects to the per
    > >> frame functional groups
    > >> 6. Add all required tags to your Enh. object still missing /empty
    > >> depending on type 1 /2

    >
    > >> The real problem is that some of the attributes have different names /
    > >> tag numbers so you need kind of a translation table. So what I refer
    > >> to as "copy all attributes" is actually a complex mapping.
    > >> I think David has a tool which does a basic transformation.

    >
    > >> Thomas- Hide quoted text -

    >
    > - Show quoted text -




  6. Re: Enhanced MR

    Hi

    Shafiq wrote:

    > Thanks for answering my doubts.
    >
    > 1. I downloaded dicom3tools, could not find a windows version of
    > dcmulti.


    Did you try compiling it under Cygwin ?

    These are Unix tools and will not compile on Windows without
    a (simulated) Unix environment.

    > 2. Can I append the pixel data of each slice one by one in final
    > target Enhanced MR. Or should i need to do any process. Looking for
    > some help on this.


    Not with dcmulti ... it requires all the single-slice image files
    to be specified on the command line. It is also an extremely crude
    tool that I originally intended for a completely different purpose,
    which I then hacked up to help me to build the NEMA demonstration
    test objects; it is by no means a complete solution and does not
    build ready to use multi-frame objects without further modification.

    Be warned that if you are looking for a solution that makes valid
    multi-frame MR objects without a considerable amount of effort and
    expertise on your part, you are not going to find one. You need
    to know a lot about DICOM but also a lot about MR as well (i.e.,
    you need to know what the attributes actually mean and how they
    are used).

    For a start, there is almost always insufficient information in
    ordinary single frame MR images to build an enhanced original
    (as opposed to derived) MR image, without at least a thorough
    understanding of various vendors' private tags, extraction of
    structured information from plain text strings (e.g. anatomy
    and contrast) and usually additional information from other sources.
    The whole point of the enhanced objects is that they make much of
    the technique information that was optional or not standardized
    in the original objects mandatory in the new ones. It is thus
    difficult for any device except the original modality that is
    running the pulse sequence and has access to all information to
    properly and completely populate a valid enhanced MR object.

    Or to put it another way, Thomas' "step 6" below is the real
    problem, not just a few different tag names, since almost
    everything technique-related is mandatory and/or has standard
    defined terms or enumerated values in the new objects.

    If you do build some objects, you will need to use the validator
    from my PixelMed Java toolkit to check them, since it is designed
    for this purpose and more thorough than dciodvfy in dicom3tools.
    Even then, it will identify only structural errors, it will not
    tell you if your new object "means" the right thing.

    What exactly is it that you are trying to do and for what purpose
    (i.e., is this training, research, testing or for a product ?).

    David

    PS. Just FYI, here is an (abbreviated) example of how I made one of
    the test images in the NEMA demonstration set using my tools; note
    the use of antodc (which extracts some of the private tag
    information from the source images), the extensive editing of
    the source single file images to populate them with things that
    I "knew" from either looking at header dumps (and which were too
    hard or not worth the effort to code into the script, antodc or
    dcmulti) or making them up if not available (OK for a demo, but
    not for clinical use), and the subsequent editing of the multi-frame
    output of dcmulti:

    include Makefile.common.mk

    DISCIMG/IMAGES/DYNMAMMO: Makefile.DYNMAMMO
    mkdir -p ${TMPDIR}
    rm -f ${TMPDIR}/*
    for i in \
    ${SRCDIR}/PhilipsDemoCDs/EV51/CD3/DICOM/IM_00307 \
    ${SRCDIR}/PhilipsDemoCDs/EV51/CD3/DICOM/IM_00308 \
    ...
    ${SRCDIR}/PhilipsDemoCDs/EV51/CD3/DICOM/IM_00473 \
    ${SRCDIR}/PhilipsDemoCDs/EV51/CD3/DICOM/IM_00474 \
    ; do \
    useacqtime=`dckey -k ContentTime "$$i" 2>&1`; \
    ${ANTODC} \
    -stamp ${STAMP} \
    -r PatientName "Breast^Dynamic" -r PatientID "MF-0000002" \
    -r PatientSex "F" -r PatientBirthDate "19500704" \
    -r PatientAge "052Y" \
    -r PatientSize "1.6" -r PatientWeight "75" \
    -r PatientComments "Tumor in right breast" \
    -d AdditionalPatientHistory \
    -r StudyID "05002" \
    -r AccessionNumber "9995002" \
    -r StudyDate "${RELEASEDATE}" \
    -r SeriesDate "${RELEASEDATE}" \
    -r ContentDate "${RELEASEDATE}" \
    -r AcquisitionDate "${RELEASEDATE}" \
    -r AcquisitionTime "$$useacqtime" \
    -r AcquisitionNumber "1" \
    -r PerformingPhysicianName "Smith^John" \
    -r ReferringPhysicianName "Thomas^Albert" \
    -r PhysicianReadingStudy "Smith^John" \
    -r OperatorName "Jones^Molly" \
    -r InstitutionName "St. Nowhere Hospital" \
    -r Manufacturer "Acme Medical Devices" \
    -r ManufacturerModelName "Super Dooper Scanner" \
    -r DeviceSerialNumber "123456" \
    -r SoftwareVersion "1.00" \
    -r StationName "CONSOLE01" \
    -r BodyPartExamined "BREAST" \
    -r ReceiveCoilName "BREAST" \
    -r TransmitCoilName "BODY" \
    -r ImageType "ORIGINAL\\PRIMARY\\PERFUSION" \
    -r AcquisitionContrast "T1" \
    -r ProtocolName "3DT1DYN+GDCORFFE" \
    -r Laterality "B" \
    -r EchoTrainLength "1" \
    -r AcquisitionMatrix "0\\512\\332\\0" \
    -r MRAcquisitionPhaseEncodingStepsOutOfPlane "28" \
    -r RepetitionTime "27" \
    -r PixelBandwidth "32000" \
    -r SAR "0.199" \
    -r RescaleType "US" \
    -d TriggerTime \
    -d ImageComments \
    -d InstanceCreationDate \
    -d InstanceCreationTime \
    -r ContrastBolusAgent "Magnevist" \
    -r ContrastBolusRoute "IV" \
    -r ContrastBolusVolume "10" \
    -r ContrastBolusIngredientConcentration "469.01" \
    -r ContrastBolusAgentAdministered "YES" \
    -r ContrastBolusAgentDetected " " \
    -r ContrastBolusAgentPhase "DYNAMIC" \
    -nodisclaimer \
    -removeprivate \
    "$$i" ${TMPDIR}/`basename "$$i"`; \
    done
    ${DCMULTI} -of $@ \
    -stamp ${STAMP} \
    -temporalposition \
    -makestack \
    -nodisclaimer \
    -removeinstanceuid \
    -dimension StackID FrameContentSequence \
    -dimension InStackPositionNumber FrameContentSequence \
    -dimension TemporalPositionIndex FrameContentSequence \
    -derivedurationfromtiming \
    -r StudyDescription "Dynamic breast,1 stack,3 dimensions,temporal position,no concat" \
    -r SeriesDescription "Coronal Dynamic Contrast" \
    -r SeriesNumber "1" \
    -r SpecificCharacterSet "ISO_IR 100" \
    -d BodyPartExamined \
    -enhancedcontrast \
    ${TMPDIR}/*
    rm -f ${TMPDIR}/*


    > On Mar 6, 4:42 pm, David Clunie wrote:
    >> Hi
    >>
    >> No, yes, dcmulti in dicom3tools (though it is incomplete), no (except
    >> under cygwin possibly), my web site.
    >>
    >> David
    >>
    >>
    >>
    >> Shafiq wrote:
    >>> Thomas,
    >>> Thanks for your guidlines. I am newbie to DICOM and have lots of basic
    >>> doubts. Some of the listed are below.
    >>> 1. Is it necessary that all the MR images in a Enhanced MR should fall
    >>> under same series.
    >>> 2. Is it necessary that all the MR images in a Enhanced MR should of
    >>> Same Dimension, For. e.g. Rows, Columns, Samples Per Pixel etc should
    >>> have same values or not.
    >>> 3. What is the name of tool DAvid have, will it run on Windows and
    >>> where can I get it?
    >>> On Mar 2, 8:29 pm, "Thomas Freier" wrote:
    >>>> On 1 Mrz., 08:54, "Shafiq" wrote:
    >>>>> Please tell me how create and Enahnced MR image form two existing MR
    >>>>> images.
    >>>> Shafiq, this is not a very easy task. Just as an outline you would
    >>>> need to:
    >>>> 1. Merge the pixel data to your new Enh object. This will be a
    >>>> multiframe format (see Part 5).
    >>>> 2. Copy identical attributes like patient name and so on to the Enh.
    >>>> object
    >>>> 3. Find attributes belonging to the multi frame functional groups and
    >>>> 3a. Copy all attributes which are identical in both source objecs to
    >>>> the share functional groups
    >>>> 3b. Copy all attributes different in both source objects to the per
    >>>> frame functional groups
    >>>> 6. Add all required tags to your Enh. object still missing /empty
    >>>> depending on type 1 /2
    >>>> The real problem is that some of the attributes have different names /
    >>>> tag numbers so you need kind of a translation table. So what I refer
    >>>> to as "copy all attributes" is actually a complex mapping.
    >>>> I think David has a tool which does a basic transformation.
    >>>> Thomas- Hide quoted text -

    >> - Show quoted text -

    >
    >


  7. Re: Enhanced MR

    David,

    > Did you try compiling it under Cygwin ?


    I don't have unix/linux...

    > What exactly is it that you are trying to do and for what purpose (i.e., is this training, research, testing or for a product ?).


    I have few MR images from Toshiba/Siemens and trying to generate
    Enahnced MR image from these images for understanding the IOD of
    Enahnced MR images. During my study, I have encountered the follwoing
    problems.

    1. Filling attributes of Image Pixel Module:

    Pixel Data:

    I assume that if all the MR slices are uncompressed just appending all
    the pixeldata of individual MR slices will produce the required pixel
    data of Enahnced MR images. Is there any order to be followed? If the
    slices are compressed can I use the same method? Is there any
    possibility of slices exist with offset table and exist with out
    offset table. If so how to organize them?

    2: Filling the attributes of Enahnced Contrast/Bolus Module:

    I assume all the attributes required for Enahnced/Contrast module will
    available in MR slices, because these are informations captured at the
    time of image acquition.

    3: Filling the attributes of Acuisition Context module:

    Since MR images does not have acuquistion module, from where this
    informations are available.

    Tx,
    Shafiq












+ Reply to Thread