Hi @joha09,
The “TissueCorr” values have been adjusted for different tissue composition (GM, WM, CSF) within the voxel, and associated differences in water concentration and relaxation times – using the method of
Gasparovic et al, 2006
The “AlphaCorr” values additionally account for differences in typical metabolite concentration between GM and WM (optionally normalised to the group average), per Harris et al, 2015
In almost all scenarios, you would want to report corrected values, not the raw water scaled ones. Any of the above corrected values could be defensible, perhaps depending on how much heterogeneity you expect within the voxel, and across subjects.
Alpha correction is only meaningful for metabolites where we have some prior knowledge about the relative concentration in GM vs WM (see Gannet alpha correction of Glx - #2 by mmikkel for some papers showing this ratio)