Abstract #0918

Quantization of ME-COSI Data with Prior Knowledge Fitting

Gaurav Verma¹, Neil Wilson², Scott Logan Lipnick², Nagarajan Rajakumar³, Michael Albert Thomas³

¹Biomedical Engineering, UCLA, Los Angeles, CA, United States; ²Biomedical Physics, UCLA, Los Angeles, CA, United States; ³Radiological Sciences, UCLA, Los Angeles, CA, United States

To quantify the 4D data generated by ME-COSI, eighteen scans of a physiological gray matter phantom were acquired. A central voxel from each acquisition was extracted and its spectrum was fitted using ProFit, a prior knowledge fitting algorithm for 2D MRS. Cramer-Rao Lower Bounds for the fit measured with ProFit were 0.3 to 16.5 for most metabolites. Across all acquisitions the coefficient of variation ranged from 2 to 21% for most metabolites. Glutamate/glutamine were overestimated possibly due to inclusion of an erroneous peak during quantization, and lactate peak showed poor fitting and reproducibility, likely due to its low concentration.