Clarification on units for tissue-corrected water-scaled metabolites in Osprey (LCModel fitting)

Analysis / Modeling
1

Hi everyone,

I am relatively new to MRS data processing and am using Osprey with LCModel fitting for short-TE unedited data. Osprey provides tissue-corrected water-scaled metabolite estimates, but I would like to confirm how to interpret the units of these values when LCModel is used for model fitting rather than Osprey’s own fitting.

My questions:

  1. When using LCModel via Osprey, do the tissue-corrected water-scaled metabolite estimates correspond to a quantitative concentration unit (e.g., molal, mmol/kg), or should they be treated as institutional units?

  2. Do I need to modify anything in the LCModel control file to ensure correct quantitative scaling, or does Osprey handle this internally?

Any clarification on unit conventions when using LCModel within Osprey, and whether LCModel control-file settings affect the final tissue-corrected values, would be greatly appreciated. If there is a recommended wording for describing this quantification approach in a Methods section, I would also be grateful.

Thanks in advance!

2

Hi @MarNil,

Thanks for your patience. Just catching up with forum posts.

Here are some thoughts:

  1. In my view, we should generally treat any attempt at tissue-corrected water-scaled estimation as institutional units. That is because the procedure is not universal. Instead, it still relies quite strongly on the assumptions made in terms of tissue-specific water densities, water relaxation times, and metabolite relaxation times. Different researchers may use different assumptions. They all try to approximate a quantity that might be expressed in molal or molar concentration units, but to me, we’re not standing on firm enough ground to be actually able to claim that we do. (FWIW, what Osprey tries to get to is mmol/kg)

  2. If you don’t provide your own LCModel control file, Osprey generates one for you from a template that overrides many of LCModel’s internal defaults. For example, if you just ran LCModel on its own, it would assume a short-TE acquisition in pure white matter, and set values for wconc, and atth2o accordingly. We set atth2o to 1 and wconc to 55 M, so that we can carry out the entire Gasparovic correction model outside of LCModel. If your own LCModel control file that you provide overrides any of these control files, this will of course affect the results.

I hope this helps - let me know if you need further explanation!

Best,
Georg

3

Thank you so much for your clarifying response, Georg!

1 Like