Consistency of frontal cortex metabolites quantified by magnetic resonance spectroscopy within overlapping small and large voxels

Abstract Single voxel magnetic resonance spectroscopy (MRS) quantifies metabolites within a specified volume of interest. MRS voxels are constrained to rectangular prism shapes. Therefore, they must define a small voxel contained within the anatomy of interest or include not of interest neighbouring tissue. When studying cortical regions without clearly demarcated boundaries, e.g. the dorsolateral prefrontal cortex (DLPFC), it is unclear how representative a larger voxel is of a smaller volume within it. To determine if a large voxel is representative of a small voxel placed within it, this study quantified total N-Acetylaspartate (tNAA), choline, glutamate, Glx (glutamate and glutamine combined), myo-inositol, and creatine in two overlapping MRS voxels in the DLPFC, a large (30×30x30 mm) and small (15×15x15 mm) voxel. Signal-to-noise ratio (SNR) and tissue type factors were specifically investigated. With water-referencing, only myo-inositol was significantly correlated between the two voxels, while all metabolites showed significant correlations with creatine-referencing. SNR had a minimal effect on the correspondence between voxels, while tissue type showed substantial influence. This study demonstrates substantial variability of metabolite estimates within the DLPFC. It suggests that when small anatomical structures are of interest, it may be valuable to spend additional acquisition time to obtain specific, localized data.