Participant-specific macromolecule information improves quantification of GABA-edited spectra at 3T

Purpose: To evaluate the effects of different MM fitting strategies on gamma-aminobutyric acid (GABA) estimates using MEGA-PRESS at 3T. Methods: Thirteen healthy individuals were scanned in the parietal lobe and centrum semiovale. GABA-edited scans were performed at TE=64ms (GABA+MMs) and TE=80ms (MM-suppressed GABA). MM-edited spectra were acquired using MEGA-PRESS with an inversion pre-pulse. MM peaks at 3.0 (MM3.0), 2.05 (MM2.05), and 0.94 (MM0.94) ppm were parameterized to form five MM basis categories. Cohort-mean/individual MM basis sets were constructed with all MM peaks combined (MMtogether), or with MM2.05 fit separately from MM0.94+MM3.0 (MM2+1). Simulated-MM bases were scaled by a participant-specific MM0.94/MM3.0 amplitude-ratio estimated from 64-224 transients. Fit quality numbers (FQNs) between the five MM parameterization methods were calculated to compare fit quality. Intraclass and Pearson's correlations were calculated to assess for agreement of GABA concentrations between the MM-suppressed GABA scan and the GABA+ scan with MMs fit out. Results: GABA concentrations derived from fitting with simulated-/individual-MM bases and MM2.05 fit separately lead to good--excellent agreements and high correlations with MM-suppressed GABA concentrations. Fit quality was significantly (51--63%) worse when fit with the MMtogether methods. GABA concentrations derived from fitting with a simulated MM basis demonstrated strong agreement with MM-suppressed GABA concentrations even when estimating the MM0.94/MM3.0 amplitude-ratio from a reduced number of transients (64--112). Conclusion: A participant-specific MM basis with MM2.05 excluded/fit separately yielded accurate GABA concentrations and superior fits. An accurate estimation of the MM0.94/MM3.0 amplitude-ratio can be achieved from a short 3-minute MM scan.