could anyone comment on whether they would discard these data? they look very noisy to me.
They are from a HERCULES scan, in a 2 x 4 x 2 voxel in ACC
The SNR of NAA in the QM matrix is 29
I agree that they look pretty noisy, and they suffer heavily from the good old out-of-voxel (OOV) artefact everywhere where an editing pulse hits. There’s no discernible GABA signal in the GABA-edited difference spectrum.
You might try a few of the other spectral registration options (no success guaranteed) - possibly RestrSpecReg, then add another parameter to the job file opts.SpecRegRange = [1.8 3.2]. This will limit the within-spectrum alignment to the range including the major singlets.
Alternatively, try deactivating spectral registration altogether (choose none). I’ve seen a few heavily OOV-contaminated datasets being rescued by that. Again, no panacea.
If none of this helps, then yes, this should be discarded (except for tNAA, tCr, tCho, mIns, etc. from the sum spectrum - these should still be fine!).
1 Like