I’d say this is progress (but not perfect, which may be hard to get to in this location).
-
These look better in terms of SNR. The one on the right has substantial lipid contamination (the large peaks between 0.5 and 2 ppm, you’ll see it even more clearly in the edit-OFF spectra) which should be avoided. You need to either move the voxel further away or change the gradient directions (more info in this thread: MEGA-PRESS Study Design Help - Study & Experiment Design - MRSHub Forum)
-
Linewidth is still not great (possible acceptable), but that’s not a surprise for very frontal locations.
-
The aligment (spectral correction) is also not optimal; there looks to be a subtraction artefact (that little dip in the 3-ppm signal in the left spectrum should not be there). You can try the other alignment options that Gannet offers.