Lattice permutation for reducing motion artifacts in radial and spiral dynamic imaging

Abstract

Radial and spiral trajectories exhibit favorable characteristics for dynamic imaging. Nevertheless, changes in image contents during acquisition lead to inconsistencies in the k-space data, which are manifested as streaks or spiral artifacts, respectively. This work proposes the concept of lattice permutation to reorder the data segments for artifact suppression. This acts to reshuffle the alias pattern along the temporal frequency axis. The proposed approach is well suited to sliding window reconstruction, although more sophisticated methods are also possible. For typical image series where the signal energies are concentrated in the low temporal frequencies, the permutation displaces most of the aliased signals from the low temporal frequencies to the high temporal frequencies, where they are attenuated by sliding window reconstruction, while the signals in the low temporal frequencies are mostly contaminated by aliasing from the much weaker signals in the higher temporal frequencies. This results in considerably reduced artifacts without any increase in scan time. In practice, lattice permutation achieves similar artifact suppression as the bit-reversed order, but with a less stringent restriction on the number of segments. At the same time, it provides a more powerful approach to controlling the alias pattern exactly. Results from real-time cardiac imaging are demonstrated.