Abstract
BACKGROUND AND PURPOSE: In in vivo magnetic resonance spectroscopy (MRS) of the brain, glycine (Gly) is traditionally separated from the overlapping signal of myo-inositol (mI) through the use of intermediate (e.g., 130-140 ms) or long (270-280 ms) echo times (TE). However, no quantitative comparisons have been performed to date comparing the performance of clinically available MRS sequences to differentiate mI and Gly as a function of TE. METHODS: In vivo spectra recorded with two clinically available MRS pulse sequences (single voxel PRESS and semi-LASER 2D-MRSI) with short (35 ms), intermediate (135 ms), and long (280 ms) echo times in a neonate with clinically suspected nonketotic hyperglycinemia were compared to those recorded from phantoms, and spectral simulations. RESULTS: In vivo spectra recorded at short and intermediate TE spectra showed signals at 3.5 ppm that could arise from either mI or Gly; however, long TE spectra showed an absence of signal in this spectral region, which was consistent with the final clinical diagnosis of hypoxic-ischemic encephalopathy. Phantom data and spectral simulations demonstrated that at intermediate TE, mI has a "pseudo-singlet" appearance that is very similar to that of Gly. CONCLUSIONS: Long echo times are used to best discriminate Gly from mI if specialized sequences and analysis methods are not available. Quantitative spectral analysis methods may also assist in correctly assigning Gly and mI.