Reducing motion artifacts in two-dimensional Fourier transform imaging.

The effects of motion in two-dimensional Fourier transform imaging (2DFT) are considered. Specific calculations describing the case of periodic motion are presented. The results predict the commonly seen artifact of image replication, sometimes referred to as ghosting. Expressions for both position and amplitude of these ghosts are derived. Simulated examples illustrate the image degradation for pulsatile flow and in plane motion. Several methods of reducing motion artifacts are then suggested. These include: randomization of views, averaging views, matching repeat times to the respiratory period, hybrid imaging, ROPE and COPE. The latter two methods reorder the data acquisition to destroy the coherence of the motion. They do not increase the data acquisition time and promise to be part of the standard approach to remove motion artifacts. The final step in actually recovering ideal resolution can be accomplished by using a model of the motion and a generalized transform inversion technique.