Feasibility of marker-free motion tracking for motion-corrected MRI and PET-MRI

Prospective motion correction is a very promising compensation approach for magnetic resonance imaging (MRI) studies impacted by motion. It has the advantage over retrospective methods of being applicable to any pulse sequence. In prospective motion correction of brain studies, the magnetic field gradients and radio frequency waveforms are adjusted in real time in response to motion of the head, thereby maintaining a fixed frame of reference for the brain inside the scanner. A key requirement of this approach is accurate and rapidly sampled head pose information. Optical motion tracking is typically used to obtain these pose estimates, however current methods are limited by the need to attach physical markers to the skin. This readily leads to decoupling of the head and marker motion, reducing the effectiveness of correction. In this work we investigate the feasibility and initial performance of an optical motion tracking method which does not require any attached markers. The method relies on detecting natural features or amplified features (from skins stamps on the forehead) using multiple cameras, and estimates pose using a 3D-2D registration between a growing database of known 3D locations on the forehead and these features. We have performed out-of-bore and in-bore experiments to test the accuracy performance of this marker-free method for very small feature patches consistent with the limited visibility afforded by head coils used during imaging. The results showed excellent agreement between the marker-free method and our current ground truth method based on wireless MR-sensitive markers