GABA spectroscopy is increasingly important at 7T, however two potential confounds have been widely ignored by the spectroscopic community: 1) Duyn et al¹ showed a potential 5-10Hz frequency shift between grey and white matter in an imaging experiment, and 2) the fact that the GABA is circa 9 times (90% GM, 10% WM)² more concentrated in grey matter. To investigate whether these confounds could affect GABA spectroscopy at 7T we combined frequency mapping with spectroscopy enabling us to compute grey and white matter lineshapes. With this information we can test whether the frequency shifts reported by Duyn et al are relevant for the larger voxels used in spectroscopy, and whether the grey matter lineshape differs from that of white matter, which should then be ideally taken into account when using analytical techniques such as the LC model approach that is based on an assumption of a single lineshape per voxel. We acquired MRI and MRS data from the occipital cortex of 9 healthy volunteers (5M/4F; 29.7 ± 3.78YO) using the Semi-LASER sequence³ and a 7T system (Siemens, Erlangen) with the following scan parameters: 8cm³ isotropic voxel, TE/TR/NEX=68ms/4500ms/64. A 3D MPRAGE was acquired as an anatomical reference before spectroscopy. B0 shimming was performed by fastest map4. A 3D GRE phase imaging acquisition was performed after the spectroscopy scan to have a frequency offset map at the same location as the spectroscopic voxel with; 1mm³ isotropic voxel with no spacing, FOV = 256 * 256 *180mm³, TEs = 10/14.24/18.48/22.72ms, FA=20°. We implemented the Dixon method⁵ to generate a frequency offset map in the spectroscopy voxel. We reconstructed a lineshape of the water signal by combining individual sub voxels of the frequency offset map. We separated grey matter and white matter regions inside the spectroscopy voxel by the histogram segmentation method based on the T1 weighted images, which were used for an anatomical reference. Grey matter and white matter lineshapes were computed by recombining frequency components at the equivalent voxel position corresponding to grey matter and white matter tissue. We excluded CSF components.We could reconstruct an identical lineshape to that of the water signal by imaging regardless of field homogeneity (Fig.1). Fig 2 shows a frequency shift and a linewidth change of grey (blue) and white (Green) matter lineshapes from water lineshape (red): water being the sum of both signals. Table 1 averages the frequency shift and the linewidth change in grey and white matter. This study has demonstrated a method using imaging to separate grey and white matter lineshape from a single spectroscopic voxel, and found a smaller level of the frequency shift than the level Duyn et al showed in an imaging experiment. This slight shift would cause no significant effect in the larger voxel used spectroscopy. Furthermore, when we consider the errors in fitting the line, the difference in the lineshape is also very slight. Hence, we conclude that we can probably safely use a single lineshape for fitting using LCModel.