Prospective motion correction for magnetic resonance spectroscopy

Proton magnetic resonance spectroscopy (1H-MRS) is an important tool in both research and clinical settings. However, 1H-MRS spectra are susceptible to artifacts and distortions due to physiological motion, which cause reduced peak amplitudes, shifts in frequency and phase, decreased quality in water suppression and change in voxel partial volume. Due to these artifacts, motion correction for 1H-MRS in the brain is desirable, especially for young subject and patients with movement disorders. Although subject motion is a common problem, little is also known about the properties of head motion in the MR environment. Using 4 healthy participants, we characterized head motion using predetermined motion patterns. Significant velocities and accelerations from head motion were recorded, especially during fast right lateral and flexion movements, of approximately 1g. Simulated tremors showed power in the 2 to 11Hz range. Kalman filtering, with the inclusion of expected motion statistics, improved involuntary motion tracking. Prospective motion correction was performed on 5 healthy subjects during Z translations and right lateral rotations. To determine changes in spectral quality, the ratio of Cho and Cr was used due to their similar sensitivity to changes in partial volume. Average change in Cho/Cr was 14.6% (+/- 3.4%) for uncorrected and 1.1% (+/- 3.3%) for corrected. The ratio of Cho/Cr, show a significant difference (t=15.5, P=0.0001) between uncorrected and corrected data. All other changes in metabolite ratios with Cr were not significant. Motion plots for both uncorrected and corrected lateral rotations showed an average rotation (Rz) of 12.5° (+/-5°), with expected X and Y translations. During uncorrected Z translations lipid peak from the skull was present in spectra however, corrected spectra were similar to the baseline. Our data demonstrate that prospective motion for 1H-MRS, using single-camera RGR tracking, significantly reduces spectral artifacts and quantitation errors due to motion. Movements that happen to introduce large lipid peaks are easily identifiable during Quality Assurance. However, other movements may change metabolite ratios in a more subtle manner that cannot be identified visually. Overall, adaptive motion correction for MRS, using single camera RGR tracking, improved spectral quality and reproducibility in subjects who move during scans.