Probabilistic spectral registration for non-edited MRS

Hi @admin and @Helge,

I have a few spectra for which robust spectral registration and restricted spectral registration did not work well, but probabilistic spectral registration worked. Is ‘ProbSpecReg’ the method that @mmikkel et al. developed for HERMES (Frequency and phase correction for multiplexed edited MRS of GABA and glutathione)? Is it okay to use this option for non-edited MRS, specifically for ultra-short TE STEAM?


Hi @Meredith,
Yes - ProbSpecReg is the method originally developed for HERMES that you are referencing. It’s fine to use this for non-edited or MEGA MRS. The approach uses the cauchy distribution to be more robust against outliers for each sub-spectrum separately. For HERMES this will include all 4, for MEGA 2, and for unedited data simply 1 group of spectra.


1 Like