Menu

Recon for dynamic dataset

Xia
2014-03-02
2014-03-10
  • Xia

    Xia - 2014-03-02

    Hi, I am trying to reconstruct a dynamic 2D data set, the settings for preparation gadget and reconstruction gadgets are as follows:

      <name>gpuRadialSensePrepGadget</name>
      <dll>gadgetron_gpuradial</dll>
      <classname>gpuRadialSensePrepGadget</classname>
      <property><name>deviceno</name><value>0</value></property>
      <property><name>mode</name><value>2</value></property>
      <property><name>profiles_per_frame</name><value>300</value></property>
      <property><name>rotations_per_reconstruction</name><value>1</value></property>
      <property><name>buffer_frames_per_rotation</name><value>0</value></property>
      <property><name>buffer_length_in_rotations</name><value>0</value></property>
      <property><name>buffer_convolution_kernel_width</name><value>5.5</value></property>
      <property><name>buffer_convolution_oversampling_factor</name><value>1.25</value></property>
      <property><name>reconstruction_os_factor_x</name><value>2</value></property>
      <property><name>reconstruction_os_factor_y</name><value>2</value></property>
    
    
    
      <name>gpuCgSenseGadget_slice0</name>
      <dll>gadgetron_gpusense</dll>
      <classname>gpuCgSenseGadget</classname>
      <property><name>pass_on_undesired_data</name><value>true</value></property>
      <property><name>deviceno</name>                <value>0</value></property>
      <property><name>sliceno</name>                 <value>0</value></property>
      <property><name>number_of_iterations</name>    <value>10</value></property>
      <property><name>cg_limit</name>                <value>1e-6</value></property>
      <property><name>oversampling_factor</name>     <value>1.25</value></property>
      <property><name>kernel_width</name>            <value>5.5</value></property>
      <property><name>kappa</name>                   <value>0.3</value></property>
    

    Basically, I just altered the value '', and kept the default value for all the other parameters. Here are my questions:

    1. This reconstruction gadget doesn't do view-sharing, does it?
    2. The recon seems to fail when I set 'profiles_per_frame' to less than 300.
    3. My total number of views is about 3450, is there anyway I could get more images out of this data set if no view-sharing is applied, please?

    Thanks a lot!

     

    Last edit: Xia 2014-03-02
    • Michael Hansen

      Michael Hansen - 2014-03-02

      What does "seem to fail" mean? Does it fail and if so, is there some kind
      of message in the log?

      On Sun, Mar 2, 2014 at 5:49 PM, Xia zxmri@users.sf.net wrote:

      Hi, I am trying to reconstruct a dynamic 2D data set, the settings for
      preparation gadget and reconstruction gadgets are as follows:

      gpuRadialSensePrepGadget
      gadgetron_gpuradial
      gpuRadialSensePrepGadget
      deviceno0
      mode2
      profiles_per_frame300
      rotations_per_reconstruction1
      buffer_frames_per_rotation0
      buffer_length_in_rotations0
      buffer_convolution_kernel_width5.5
      buffer_convolution_oversampling_factor1.25
      reconstruction_os_factor_x2
      reconstruction_os_factor_y2

      <gadget>
      <name>gpuCgSenseGadget_slice0</name>
      <dll>gadgetron_gpusense</dll>
      <classname>gpuCgSenseGadget</classname>
      <property><name>pass_on_undesired_data</name><value>true</value></property>
      <property><name>deviceno</name> <value>0</value></property>
      <property><name>sliceno</name> <value>0</value></property>
      <property><name>number_of_iterations</name> <value>10</value></property>
      <property><name>cg_limit</name> <value>1e-6</value></property>
      <property><name>oversampling_factor</name> <value>1.25</value></property>
      <property><name>kernel_width</name> <value>5.5</value></property>
      <property><name>kappa</name> <value>0.3</value></property></gadget>

      And I have a couple of questions:

      1. This reconstruction gadget doesn't do view-sharing, does it?
      2. The recon seems to fail when I set 'profiles_per_frame' to less
        than 300.
      3. My total number of views is about 3450, is there anyway I could get
        more images out of this dataset if no view-sharing is applied, please?

      Thanks a lot!

      Recon for dynamic datasethttps://sourceforge.net/p/gadgetron/discussion/general/thread/d57fdf63/?limit=25#bde6

      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 - 2014-03-03

        Dr. Hansen,

        "Seems to fail" means, the reconstruction does go through and give me a .h5 file, however, when I open the images (there will be about 10 frames) with 'hdfview', all the images are totally black.

        Oh, I've got another question today. I found out that there is a 'sliding_window_profiles' property for gpuRadialSensePrepGadget. I tried to use it, but it doesn't work. Every time I add this property into my .xml file, I got the following message in client window:

        (28690|140195003692800) Unable to send acquisition message identifier
        (28690|140195003692800) Failed to write message to Gadgetron

        And the reconstruction returns to normal after removing it. Do you have any idea about what is the problem, please?

        Thanks

        Xia

         
        • Michael Hansen

          Michael Hansen - 2014-03-03

          We need a lot more information in order to assist you with this. The error messages from the client are not helpful, we need the log on the Gadgetron side. Most likely we will also need a dataset (you can upload to Dropbox or something like that) and the exact command you wrote and the XML configuration file that you used. There is really no way to comment on the issues based on the information provided.


          Sent from a mobile device - please excuse brevity and/or typos.

          On Mar 3, 2014, at 5:08 PM, "Xia" zxmri@users.sf.net wrote:

          Dr. Hansen,

          "Seems to fail" means, the reconstruction does go through and give me a .h5 file, however, when I open the images (there will be about 10 frames) with 'hdfview', all the images are totally black.

          Oh, I've got another question today. I found out that there is a 'sliding_window_profiles' property for gpuRadialSensePrepGadget. I tried to use it, but it doesn't work. Every time I add this property into my .xml file, I got the following message in client window:

          (28690|140195003692800) Unable to send acquisition message identifier
          (28690|140195003692800) Failed to write message to Gadgetron

          And the reconstruction returns to normal after removing it. Do you have any idea about what is the problem, please?

          Thanks

          Xia

          Recon for dynamic dataset

          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 - 2014-03-10

            Dr. Hansen

            Sorry for the delayed reply.

            We tried to run the sliding_window reconstruction on our colleague's computer, and it succeeded. So it probably has to do with either the gadgetron version we are using, or the hardware of our computer. I will check that out, and let you know if we find something.

            Thank you all the same.

            Xia

             
MongoDB Logo MongoDB