Menu

Recon for hybrid 3D k-space data with partial acquisition

Xia
2014-02-16
2014-03-02
  • Xia

    Xia - 2014-02-16

    Hi, I am having some problems reconstructing my MRI data with Gadgetron. Here is a brief description of my data:

    It is a set of dynamic data acquired in hybrid 3D k-space----phase-encoding in the slice direction (but 80% partial acquisition: 26 phase-encoding steps for 32 slices) and golden angle radial spokes in x-y plane.

    As far as I understand, the built-in functionality of Gadgetron deals with 2D data as well as multi-slice data, but not this type.

    So I am wondering what I can do to get it reconstructed using Gadgetron, please?

    Thanks a lot!

     
    • Michael Hansen

      Michael Hansen - 2014-02-16

      It is basically impossible to answer this question without more knowledge
      about your data.

      Do you acquire the same k-space lines in each of your slice phase encoding
      steps?

      If yes, then you can simply to a point by point iFFT along the slice
      direction to and then send the data onto the 2D golden angle radial recon
      slice by slice. You would need to write some code for those initial steps
      and book keeping. Again, hard to say in general what to do since I don't
      know your data layout.

      If no, then you need to treat it as a 3D encoding space. There would be
      some work to do to make that work and depending on the size of your problem
      it may not be suitable for GPU processing which is what is currently done
      in the golden angle recon.

      Michael

      On Sun, Feb 16, 2014 at 11:11 AM, Xia zxmri@users.sf.net wrote:

      Hi, I am having some problems reconstructing my MRI data with Gadgetron.
      Here is a brief description of my data:

      It is a set of dynamic data acquired in hybrid 3D
      k-space----phase-encoding in the slice direction (but 80% partial
      acquisition: 26 phase-encoding steps for 32 slices) and golden angle radial
      spokes in x-y plane.

      As far as I understand, the built-in functionality of Gadgetron deals with
      2D data as well as multi-slice data, but not this type.

      So I am wondering what I can do to get it reconstructed using Gadgetron,
      please?

      Thanks a lot!

      Recon for hybrid 3D k-space data with partial acquisitionhttps://sourceforge.net/p/gadgetron/discussion/general/thread/1567e5ab/?limit=25#fcce

      Sent from sourceforge.net because you indicated interest in
      https://sourceforge.net/p/gadgetron/discussion/general/

      To unsubscribe from further messages, please visit
      https://sourceforge.net/auth/subscriptions/

       
      • Xia

        Xia - 2014-02-17
         

        Last edit: Xia 2014-02-17
      • Xia

        Xia - 2014-02-17
         

        Last edit: Xia 2014-02-17
      • Xia

        Xia - 2014-02-17

        Dr. Hansen,

        Hi, I appreciate your kind response. Yes, as you expected, we acquired the same radial spoke for all the 26 slice phase encodes, and then move to the next angular position. We tried to do what you have suggested: do some programming to first resolve all the slices and feed the data into Gadgetron. And the most convenient way for me to do that is using Matlab, so I resorted to the example Matlab code "test_create_dataset.m" under the ismrmrd/example/matlab directory, trying to use it as a template for writing my own code. Unfortunately, that example code doesn't seem to work. I am sure I have the newest ISMRMRD and have all the necessary libraries installed. So do you have any suggestions for this, please?

        Thank you very much for your time.

        Regards
        Xia

         
        • Michael Hansen

          Michael Hansen - 2014-02-17

          As you can see in other threads on this forum, the Matlab components are an
          undocumented, buggy part of the Gadgetron, which is still being developed.
          I would not recommend that as a place to start right now.

          Your best bet is to create a Gadget (in C++), which a) buffers the incoming
          projections, b) when all slice-phase encoding steps have been completed
          does an FFT along the slice direction, c) sends the profiles down the
          stream. There is really no other way of doing this than just rolling up
          your sleeves and programming it. It can probably be done in 50 lines or
          less, but it may take you some time the first time around.

          Before going down that path, I would experiment with the 2D golden angle
          reconstructions first and see if they actually do what you would need them
          too.

          On Mon, Feb 17, 2014 at 11:52 AM, Xia zxmri@users.sf.net wrote:

          Dr. Hansen,

          Hi, I appreciate your kind response. Yes, as you expected, we acquired the
          same radial spoke for all the 26 slice phase encodes, and then move to the
          next angular position. We tried to do what you have suggested: do some
          programming to first resolve all the slices and feed the data into
          Gadgetron. And the most convenient way for me to do that is using Matlab,
          so I resorted to the example Matlab code "test_create_dataset.m" under the
          ismrmrd/example/matlab directory, trying to use it as a template for
          writing my own code. Unfortunately, that example code doesn't seem to work.
          I am sure I have the newest ISMRMRD and have all the necessary libraries
          installed. So do you have any suggestions for this, please?

          Thank you very much for your time.

          Regards
          Xia


          Recon for hybrid 3D k-space data with partial acquisitionhttps://sourceforge.net/p/gadgetron/discussion/general/thread/1567e5ab/?limit=25#fcce/01f0/5d82

          Sent from sourceforge.net because you indicated interest in
          https://sourceforge.net/p/gadgetron/discussion/general/

          To unsubscribe from further messages, please visit
          https://sourceforge.net/auth/subscriptions/

           
          • Xia

            Xia - 2014-03-02

            Dr. Hansen,

            I have found a way around this problem and successfully fed my data into Gadgetron, and the recon looks good.

            Thanks again for your kind help!

            Regards
            Xia

             
MongoDB Logo MongoDB