In the United States, traumatic brain injury (TBI) is the number one cause of neurological disability in young adults. Most cases are classified as mild TBI (mTBI), which can result in chronic post-concussive symptoms (PCS). Clinical management of mTBI patients, unfortunately is rarely based on biomarkers, due to the lack of findings on conventional imaging. Fortunately, quantitative MR techniques, such as diffusion tensor imaging (DTI) and proton MR spectroscopy (¹H-MRS) have shown microstructural and biochemical changes in normal-appearing tissue, motivating research into their potential clinical use^1,2. Since axonal injury is the primary TBI outcome, and has been extensively documented with DTI^1,3 and ¹H-MRS^{2, 4-7}, our goal was to characterize its regional distribution in order to identify regions prone to disproportionate injury, hence, candidate targets for single-voxel or multi-voxel clinical ¹H-MRS of mTBI. We utilized a dataset of PCS-positive patients in which we previously showed diffuse axonal abnormalities, i.e. lower global white matter (WM) N-acetyl-aspartate (NAA) levels compared to controls⁸.Subjects: Fifteen PCS-positive mTBI patients (Table 1) and 12 age- and gender-matched controls. Data acquisition: T1-weighted MRI (MP-RAGE), T2-weighted MRI (FLAIR), B₀ shimming, 10×8×4.5 cm (AP×LR×IS)=360 cm^{3 1}H-MRS VOI (PRESS TE/TR=35/1800 ms), encoded to 480 voxels, each 1.0×1.0×0.75 cm³ (Fig. 1). Metabolite quantification: Absolute amounts of NAA, creatine (Cr), choline (Cho) and myo-inositol (mI) were obtained using phantom replacement with correction for T1 and T2 relaxation time differences between in vivo and in vitro. Segmentation: Six WM regions (body, genu, splenium of the corpus callosum, corona radiata, frontal and occipital WM) were manually outlined in 3D on each patient's axial MP-RAGE images based on DTI atlas parcellations⁹, as shown in Fig. 1 and 2. Global WM, gray matter (GM) and cerebro-spinal fluid (CSF) masks were obtained from the MP-RAGE using SPM. Regional ¹H-MRS: All masks were co-registered with the ¹H-MRS matrix, yielding their volume in every ¹H-MRS voxel. This enabled CSF and GM partial volume control: only voxels with <30% of each were retained for subsequent analyses. Next, the goal was to ensure similar statistical precision for the analysis of each WM region, since larger regions (e.g. corona radiata) would normally contribute more voxels than smaller regions (e.g. genu), which would translate into better precision for the former because of voxel averaging. To avoid this scenario, we used similar number of voxels for each WM region by determining the optimal trade-off between mask inclusion and number of selected voxels. The resulting minimal voxel mask fractions used for each region were 40% for body, 50% for genu and splenium, 60% for occipital and 70% for corona radiata. Finally, only voxels with metabolite Cramer-Rao lower bounds<20% and 4<linewidths<13 Hz were included. Statistics: Unpaired student t tests with Bonferroni correction for multiple comparisons.The metabolic concentrations in each WM region are shown in Table 2, and their distributions are plotted in Fig. 3. There were no statistically significant differences in any metabolite between patients and controls, in any WM region, even before correction for multiple comparisons. However, as evident from Fig. 3, the median NAA concentration in each WM region was lower than controls’.The WM regions were selected on the basis of being the most commonly injured, based on previous ¹H-MRS^4-7, DTI³ and histopathology^10-12. The additional criterion was that small tracts, unsuitable for the course spatial resolution of ¹H-MRS were not considered. We ensured similar statistical precision among all comparisons, albeit at the cost of accuracy; however, this was an appropriate trade-off for this study, which had the goal of comparing the ability of ¹H-MRS to detect injury amongst different regions. There were two main findings. First, while there were no significant differences, NAA in patients was lower across all WM regions. This indicated that metabolic axonal injury is truly diffuse, i.e. no region is spared, but also no region sustains injury of much larger proportion than others. Second, despite careful partial volume control, regional analysis of multi-voxel data lacked the sensitivity to detect (the clearly present) regional injury. This motivates the use of: (i) global WM multi-voxel approaches, which benefit from increased sensitivity (i.e. precision) due to voxel averaging¹³; or (ii) large volumes-of-interest (e.g. single-voxels) in pure WM, without consideration of locale.Metabolic axonal injury in mTBI is diffusely distributed among commonly injured tracts, but multi-voxel ¹H-MRS may lack sensitivity for its detection on a regional basis. These results motivate the use of ¹H-MRS approaches with higher sensitivity, such as global WM averaging, or large single-voxels in areas of pure WM, irrespective of placement location.