• Advanced navigator methods can enable the rapid tracking and correction of motion in two or three dimensions.

• Self-navigation methods can be applied in some cases to correct for motion using the imaging data themselves, without the need for separate motion-tracking sequences.

• Multi-shot diffusion sequences incorporating self-navigated phase correction show promise for improving spatial resolution and reducing image distortion without the introduction of ghosting artifacts.

Primarily practitioners of body, cardiac, pediatric, and neuro MRI, or anyone challenged by motion problems in MR.

Upon completion of this course, participants should be able to:

• describe the effects of various types of motion on MR images;

• choose those simple motion correction measures which are most appropriate in different circumstances;

• describe new emerging motion-correction techniques for different body regions/subspecialties.

MRI’s biggest weakness is arguably its long scan times, which result in increased vulnerability to motion artifacts, producing degraded image quality, more complex patient workflow, and the need in some cases for patient sedation, restraint, or rescanning (1). A number of technologies have been developed or are under investigation to ameliorate this problem, with a view towards unlocking the full potential of MRI.

While advanced motion correction techniques are a major area of research in MRI, many other more conventional methods are currently available for implementation on most commercial scanners. These include signal averaging to cancel ghosting, use of saturation bands to suppress regions of moving anatomy, exclusion of signals from RF coil elements near moving anatomy, gradient-moment nulling to avoid artifacts from pulsatile or complex flow, swapping of phase and frequency directions to move ghost artifacts away from anatomy of interest, motion insensitive k-space trajectories (2), and use of real-time or single-shot pulse sequences to freeze motion (3). Other conventional methods incorporate motion tracking of some kind. These include use of signals from ECG, finger-plethysmographic, or respiratory-belt sensors to trigger or gate MR image acquisition to cardiac or respiratory motion. View ordering can be arranged to push motion-induced ghosts outside the anatomy of interest (4-6). And respiratory navigator sequences can be interleaved with image acquisition to track motion of the diaphragm for respiratory gating or triggering (7). A number of advanced approaches are also under development that can potentially provide significantly improved motion tracking and correction. New kinds of navigator sequences are being explored, including orbital (8), spherical (9), cloverleaf (10), and three-plane (11). These all expand the range of available motion data beyond the 1D data collected from classic navigator pencils (7), and thus enable potentially better tracking of body or head motion. A number of self-navigated pulse sequences (12-17) have also been developed which allow motion tracking from the imaging data themselves, without the need to interleave separate tracking acquisitions. These sequences include one that rotates a Cartesian “propeller” in k space (12), or another that spins through spiral interleaves (13) - for coronary artery imaging. More recent self-navigated 3D techniques have incorporated golden angle sampling in the readout (16) or phase-encode (17) dimensions, to allow quasi-uniform k-space coverage even with relatively sparse sampling.

Various groups have been developing data-driven methods to retrospectively correct for motion after data collection is complete. In fetal MRI, for instance, intersection-based motion correction has been used to retrospectively realign slices acquired at sometimes very different head positions into a 3D reconstruction (18). For non rigid-body motion, a motion model can be incorporated into the image reconstruction based on prior knowledge e.g. from navigators – The motion model and the image are refined together in each stage of an iterative reconstruction (19).

Diffusion weighted imaging (DWI) – with its extensions – is an important tool for probing tissue microstructure that is especially sensitive to motion. This is typically performed using a single-shot echo-planar imaging (EPI) sequence, in order to avoid severe ghosting artifacts that can result from inconsistent phase terms caused by physiological motion (e.g. brain pulsation) during diffusion encoding. However, single-shot EPI can result in low spatial resolution and high image distortion in areas of high B0 inhomogeneity. A number of groups have developed multi-shot (e.g. interleaved EPI or interleaved spiral) techniques that account for shot-to-shot variation in the phase error. These typically use a parallel-imaging recon step to generate a low-resolution phase map for each shot, followed by a reconstruction that integrates phase correction into a parallel-imaging algorithm to produce relatively artifact-free images (20-22).

There are a number of complicating issues that motion correction techniques sometimes must contend with, including spin history; non-uniform RF, B0, and gradient fields; through- plane motion; and directional encoding (e.g., in the case of phase-contrast or diffusion imaging). Those investigational techniques that can navigate these issues while proving robust and easy to use should be most successful in moving into routine clinical practice. The successful application of new motion correction techniques will allow MRI to expand into new areas and provide even more clinical impact.