August 15, 2016

    We are excited to announce that we have re-opened submissions to the Phase II scoring system to the public! We are doing this to encourage further work on these important problems, and see contestants push the envelope even further! Please follow the Phase II instructions on the Details page to prepare and submit your solutions. Submissions with scores better than the final winning Phase II scores (see the Results page) will appear on the Leaderboard. Good luck! Note that you must register a team and login to access the submission system in order to submit - see the bottom of the Home page to register.

November 3, 2015

  • We are happy to announce that the scoring system is live and we are now accepting submissions for this year's Challenge!
  • To submit your candidate pulses for parallel transmit and simultaenous multislice imaging, please see the 'Submit' link at the top of this page (visible to registered and logged in users).
  • Note that you must register a team and login to access the submission system in order to submit.

October 1, 2015

Today was the official launch of the Challenge! We kicked it off with a Virtual Journal Club meeting, the video of which can be viewed here: Virtual Journal Club Video. You can get the Powerpoint slides that we showed here: VirtualJournalPresentation.pptx.

Here are some of the questions we received during the meeting, and our answers:

  • How do we submit?
    Upload buttons are on the 'Submit' page of this site, which you can access once you have registered a team.
    You may upload a submission for either parallel transmit, simultaneous multi-slice RF pulse, or both. Note that you can only make one submission per hour, whether you submit only one, or both candidate pulses.
  • How do we register a team?
    You can register a team by clicking on the following link and filling out the form: Create a new account
  • When will the material be online?
    All the materials needed to get started are on the 'Details' page of this site now!
  • Will there be a leaderboard?
    Absolutely! There are actually two leaderboards, one for the pTX design and one for the SMS design:
    pTX Leaderboard, SMS Leaderboard
  • What will the scoreboard look like?
    In Phase I, the scoreboard will list the teams and their scores in increasing order, since lower scores (pulse durations) are better. In Phase II, the cases and scores will be somehow obscured to keep the cases blinded, but they will still be numerical and a ranking will be clear.
  • Will there be off-resonance maps for the parallel transmit challenge?
    No. We certainly appreciate the importance of accounting for off-resonance in multidimensional parallel transmit, but there are two reasons we have chosen not to incorporate them into the challenge. First, since the objective of the challenge is to achieve short pulse durations, the best solutions will be intrinsically more robust to off-resonance. The second reason is practical: it is simply difficult to generate an off-resonance map for the simulated cases, and it is computationally quite expensive for us to incorporate off-resonance into our profile calculations.
  • What is the peak B1 constraint for the shortest MB pulse?
    The peak B1 constraint will be 18 microTesla for both the TSE and diffusion cases in Phase I. This and other specific constraints can be found in the parameter scripts included with the code packages.
  • Do we define our own dwell time?
    Yes, you do. Your choice of dwell time will depend on balancing the total number of samples in your pulses (a total length of 20,000 samples is enforced in both cases), with the accuracy of the Bloch simulation and pattern calculations. It will probably be best to use dwell times no larger than 4 us or so, especially for the simultaneous multislice challenge.
  • When will the next virtual meeting be?
    We don't have a specific plan for the next meeting, though there will certainly be one to launch Phase II.
  • Is there extra credit if the pulses are adiabatic?
    Good one! We will be very impressed if you could win the parallel transmit with a slice-selective adiabatic pulse. In all seriousness though, since we are assuming small-tip-angle excitation in the scoring code, an adiabatic pulse would not be accurately modeled.
  • How many teams will make it to Phase II?
    We expect that the 5-10 top teams from Phase I will make it to Phase II. We don't expect to be able to admit more than that since we expect that we will need to provide some support to each of the teams to get their code submissions running on our side.
  • Is there any criteria for computation time for designing the pulse?
    Not in Phase I! Spend as much computational power as you like. In Phase II we will set some reasonable upper limit on compute time, but our advice is to not worry about it for Phase I.
  • Will the scripts work in Octave?
    We have not tested them in Octave, but with the exception of the Gmri object used to evaluate the parallel transmit patterns, all code is plain MATLAB, so it should be possible to get it all working. If you choose to do it, please let us know how it goes so we can share with others!
  • How many people can be in the team?
    As many as you like! Of course, we will ask for a single representative to take any correspondence or accept the prize if you win.
  • When is the deadline for the Phase I submission?
    The deadline will be around the end of February, but definitely get some scores up well before then! We will be in heavy communication with contestants near the end of Phase I.
  • What is the prize?
    Fame and fortune await the winner, of course! But seriously, we are looking into it! One thing we think is very important is that the winners be documented in the literature, so we will be authoring a Letter to the Editor that we will submit to MRM, that will discuss the goals, structure, and outcomes of the Challenge. Winners will then be able to cite this paper as proof.
  • Why are only VOP's provided for the parallel transmit challenge?
    We in fact have abused the term ‘VOP’ here, since we derived these matrices using the VOP clustering algorithm, but we did not perform the majorizing step. So these matrices are the B* matrices in the Eichfelder paper, and correspond exactly to specific spatial locations. That of course introduces the possibility of underestimation of true peak local SAR, but since all contestants’ entries are validated using the same matrices, the fairness of the competition is maintained, while limiting computational complexity of the scoring system.