Sarah Andrea Wijtenburg1,2, Jack Knight-Scott1,2
1Radiology, Children's Healthcare of Atlanta, Atlanta, GA, USA; 2Biomedical Engineering, University of Virginia, Charlo ttesville, VA, USA
Here, we examine the effect of SNR on the reliability of the frequency-domain LCModel and time-domain QUEST quantification methods for spectra from the human brain. 10 ms TE 1H MRS data were collected from healthy adults on a 1.5 T Siemens Magnetom Sonata MRI system. NAA, Glx, Cre, Cho, and mI were identified, and the SNR calculated for each metabolite in the time and the frequency domain. Based on the stability of the concentration values, these results indicate that NAA and Cr are best fit with LCModel, while Glx, Cho, and mI are best fit with QUEST.