3D planes-on-a-paddlewheel (POP) echo-planar imaging (EPI) is an effective non-Cartesian readout scheme realized by rotating conventional EPI readout planes about the phase encoding axis. Navigator based phase correction schemes are typically employed to account for gradient timing errors, associated trajectory errors and artifacts. In this work, we propose to use “Self Consistency for an Iterative Trajectory Adjustment” SCITA for an improved and purely data-driven removal of trajectory misalignment artifacts. As the actual k-space trajectory is derived from the imaging data, navigator acquisitions can be omitted and echo, repetition and acquisition times may be considerably shortened.
3D planes-on-a-paddlewheel (POP) echo-planar imaging (EPI) is a non-Cartesian readout scheme realized by rotating conventional EPI readout planes about the phase encoding axis (Fig. 1). The technique is effective for distortion minimization and was recently employed for whole-brain quantitative susceptibility mapping (QSM) at ultra-high field1.
Gradient timing errors are known to cause severe artifacts in EPI. Phase correction algorithms2, based on additional navigator echoes, typically have to be employed to remove the Nyquist ghosting, which arises from the shift of odd and even echoes with respect to each other along the phase encoding axis in k-space. In 3D POP EPI, an additional misalignment of the projections within the radial readout planes is introduced, which is typically neglected by navigator-based phase correction algorithms.
In this work, we propose to exploit “Self Consistency for an Iterative Trajectory Adjustment” (SCITA)3 for a purely data-driven removal of trajectory misalignment artifacts in 3D POP EPI. As the additional acquisition of navigator echoes can be omitted, the approach may facilitate shorter echo, repetition and acquisition times.
SCITA exploits the inherent oversampling of central k-space in radial imaging and obtains an estimate of the actual k-space trajectory that can be finally used for gridding. To that end, the trajectory correction technique iteratively maximizes the data consistency across the individual projections in the oversampled part of k-space by shifting individual projections with respect to the rest of the acquired k-space data by means of the GRAPPA operator4.
Measurements were performed on a non-commercial 7 T whole-body research MRI scanner (Siemens Healthcare, Erlangen, Germany) under institutional review board permission. The system was equipped with a Tx/32 channel Rx head array (Nova Medical, USA) and 3rd order shimming was employed for all measurements. One healthy 34-year-old volunteer was examined, after obtaining written informed consent and approval of the local ethics committee, and 3D POP EPI images were obtained at 1.5 mm isotropic spatial resolution using following acquisition parameters: FOV = 170 x 170 x 37 mm3, 172 projections with linearly increasing azimuthal angle within the interval [0;π[, TR = 36 ms, TE = 19 ms, ES = 1.03 ms, flip angle = 12°, ramp sampling = 6.4%. At the beginning of each readout train, 3 additional navigator echoes were acquired. All images were reconstructed offline using Matlab (The Mathworks, Natick, USA). The non-uniform fast Fourier transform (NUFFT) framework5 was used for gridding, followed by a Fourier transformation along the phase encoding axis and sum-of-squares coil combination.
To test the performance of SCITA, images were reconstructed
For the latter reconstruction, the trajectory estimates were obtained from the radial plane containing the echo and the adjacent radial plane (phase encoding step 12). To achieve a proper alignment of those two trajectories in the separate odd/even reconstruction, each trajectory was centered by iteratively maximizing the sum-of-squares k-space DC signal at k = 0 upon the application of multiple linear phase ramps in image space.
In all cases, SCITA was employed iteratively as described elsewhere3 with continuously decreasing shift lengths of 0.33, 0.15, 0.1, 0.07, 0.03 and 0.01 Δk with 20 iterations per shift length.