Siemens phantom + Hercules test showing odd concentrations


we tried running the Hercules sequence on a spherical Siemens phantom that contained a 0.1 M solution of Sodium Acetate (8.2g) and Lithium Lactate (9.6g).

we were surprised to see concentrations for metabolites in the osprey output that weren’t in the phantom (eg. non-zero estimates for GABA & Gln). This concerns us - is it correct that this unexpected behavior? does anyone have insight on what could be causing this false concentrations? are there any suggestions for how to resolve this or further tests to do?

link to all figures output by Osprey (incl. fit, load, process):

Hello sdw,

Was it just sodium acetate and lithium lactate in the phantom?

By default, Osprey assumes an in-vivo brain spectrum, and so the default set of basis functions used for the linear combination modeling contains compounds specific to that environment. It seems that Osprey has assigned some of the signal from your MR spectra to those basis functions, likely due to a slight peak overlap. I’d argue that it is expected behavior, but that the underlying model assumptions are not correct.

Can I ask what your goal is? If you want to quantify sodium acetate and lithium lactate, for example, you’ll need to simulate a basis set that contains these compounds, or perhaps model your spectra differently.

Best wishes,

thanks!! yes, upon further reflection it was nonsensical to try to use those basis functions and hope that the non-present metabolites wouldn’t be assigned any high concentrations, thanks for pointing this out : )

1 Like

Hi sdw,
when you look at your spectrum before alignment (Pre-alignment), you have a spectrum with your lactate and acetate peaks. You have measured the spectrum you want. As Christopher said, Osprey works with some prior knowledge and assumes it’s an in-vivo spectrum and I guess it uses e.g. NAA, Cr or Choline for frequency and phase correcture. Perhaps Osprey has incorrectly assigned residual water to these signals and has performed corresponding frequency correction…