Markerless high-frequency prospective motion correction for brain MRI

Head motion during MRI of the brain is widely recognized as a major
problem in both clinical practice and neuroimaging research. Small
movements can easily cause artifacts when a single image is encoded
over several minutes. If head motion can be measured during the scan
then corrections can be applied retrospectively to the k-space data,
or if the tracking information is available quickly, the image
encoding can be adjusted in real-time to compensate for head motion,
so that high-quality images are available immediately after the scan.

This talk will: 1) show how head motion can be measured with the MR
scanner or with cameras for prospective correction; and 2) present
recent results using a commercial "markerless" face tracking system
(TracInnovations, Denmark). External camera systems offer the benefits
of being independent of the scan and providing high-frequency motion
information, however, usually some form of optical marker needs to be
attached to the patient's head. The "markerless" 3D surface tracking
technology is unique in that a marker is not required, making it
highly relevant from a workflow perspective, e.g. for use with young
children. We will demonstrate markerless prospective motion
correction in high-isotropic-resolution anatomical scans as well as
widely-used clinical T2 and FLAIR MRI. Advantages of high-frequency
correction every 50 ms during continuous head motion will also be
shown.