2012 Challenge Overview

The 2012 ISMRM Challenge is on water-fat reconstruction.

Current MRI techniques for separating water and fat signals tend to work spectacularly most of the time, but they can also fail on a regular basis. These failures may lead to large water/fat swaps that are often obvious, but can sometimes be cryptic and potentially detrimental to the clinical interpretation of images.

We believe that this problem is solvable. Doing so will make clinical MRI more robust in a variety of applications (e.g. liver, heart) and in calibration (e.g. field mapping).

We want to engage the MRI community to develop the most robust water-fat reconstruction algorithm that:
  • Can generate water and fat images accurately for a wide variety of clinically relevant cases
  • Does not require manual adjustment on a case-by-case basis
  • Reconstructs within practical time limits (< 30 min per case on commodity computing hardware. e.g. PC, Mac with or without GPU)
  • We provide you with real-life cases of multi-echo MRI data from variety of anatomies at 1.5 or 3T. You apply your technique to these cases, and submit separated water and fat images. We determine a score based on the accuracy of your separation results. In February 2013, we invite the top teams to submit their algorithms, and we run them through an expanded set of cases without any case-by-case adjustments. The processing time is verified to be under <30min per case, and a physician panel verifies the separation results.

    The winner is announced at ISMRM in Salt Lake City. To earn the bragging rights, you need to either fully disclose your algorithm or share your algorithm in a black-box manner for any validation request within a year from the end of the Challenge. This is quite an achievement, if you can beat out others across a wide variety of clinical relevant cases.

    We are willing to help you out! The Challenge is arranged together with organizers from the 2012 ISMRM fat/water separation workshop, which has recently released a toolbox of current algorithms. You are more than welcome to use one or more of them as starting points, as long as you let the original authors know.