High-Temporal Resolution Direct Motion Parameter Estimation via Inertial Sensors

Motivation: High temporal resolution is crucial for accurately characterizing motion to correct MR images; however, no strategy currently offers this capability. Goal(s): Our goal was to develop a high-temporal-resolution system that directly detects the onset of motion and estimates 3 rotational parameters of a body that undergoes 6-DOF rigid motion. Approach: We developed algorithms to detect motion and estimate rotation with a 2000 Hz inertial system, while directly estimating translation via k-space cross-correlation. MR images were corrected using 6-DOF motion parameters. Tests at 3T included phantom and volunteers. Results: The proposed system significantly reduced motion artifacts and enhanced image quality compared to non-motion-corrected images. Impact: We developed computationally cheap novel algorithms that utilize a low-cost wearable inertial sensor to deliver accurate motion information with very high temporal resolution without requiring modifications to the sequence.