• prospective motion correction;
  • retrospective motion correction;
  • Kalman filtering;
  • tracking accuracy;
  • real-time;
  • conjugate gradient method


Prospective motion correction can prevent motion artifacts in magnetic resonance imaging of the brain. However, for high-resolution imaging, the technique relies on precise tracking of head motion. This precision is often limited by tracking noise, which leads to residual errors in the prospectively-corrected k-space data and artifacts in the image. This work shows that it is possible to estimate these tracking errors, and hence the true k-space sample locations, by applying a two-sided filter to the tracking data after imaging. A conjugate gradient reconstruction is compared to gridding as a means of using this information to retrospectively correct for the effects of the residual errors. Magn Reson Med, 2011. © 2011 Wiley-Liss, Inc.