head motion

Retrospective correction of head motion using measurements from an electromagnetic tracker

Paper link

  • What sequence?
    • T1-weighted MPRAGE sequence
  • What recon?
    • Each k-space line from each coil -> corrected for phase ramps -> regridded in 3D k-space (NUFFT) with the estimated rigid body motion param as suggested by Gallichan [23]
    • Motion-corrected images were reconed for each coil -> the adaptive combine method was used to create the final image.
  • What Electromagnetic tracking data look like?
    • Translation and rotation of each scan..
  • How do we get the reference image?
    • Still images (include involuntary movements)

  • What is the protocol?

    • Phantom experiments: a pineapple -> 8 times(after movements) * 45s

    • Volunteer experiments: 6 volunteers, each volunteer -> 4 volumes

      • still -> used as no motion reference image
      • four movements (random position)
      • nodding up down
      • continuous head nodding: up-down, down-up every 30 seconds (not explained too much)
        • Patient scans: eight pediatric patients (Not explain too much)

    • Metrics:

      • normalized root-mean-square error (NRMSE) and the structural similarity (SSIM) index

        (using FSL tool BET [26]) relative to this motion-free reference image.

      • a reference-free metric, average edge strength (AES) was calculated for each scan

        As suggested by Aksoy et al. [27], AES values were normalized by the corresponding slice in the “no motion” dataset.

  • Limitation: only in-plane motion, but inability of correction for though-plane motion in 2D sequence.