I’m analyzing the brain metabolites for mice brain regions (Hippocampus, Thalamus and Cortex) using LC Model GUI. the problem that I’m having is, when I used short TE(18ms) the absolute concentrations were in range similar to literature online however after using long TE(135ms) the absolute concentration shoots 4 to 5 times higher as compared to short TE. In both cases respective basis sets were used for 7T.
Could you please guide me and let me know what am I missing here ? Is there anything to add in control parameters especially for long TE to quantify metabolites ?
Basically, it multiplies the estimated metabolite model amplitudes and the estimated water signal area with two scalars that contain the necessary relaxation corrections. The kicker is, these scalars are not calculated from TE automatically, but they are instead input as control file parameters.
By default, the value for the parameter atth2o is 0.7, which is about what you get for T_2 relaxation at TE = 18 ms if T_2 of water is 50 ms (e^{-(TE/T_2)} = e^{-(18/50)} ≈ 0.69). In contrast, the default value for attmet is 1.0, because metabolites tend to have longer T_2 than water, and the exponent becomes very small at short echo times.
TLDR, the default relaxation correction LCModel uses for water and metabolites is only good for short-TE, but becomes invalid when you go to longer TE. So what you need to do is adjust atth2o and attmet if you go to the longer echo time.
Also note that LCModel assumes a default water concentration of ~36M (set in the control parameter wconc, that value corresponds to pure white matter) and has no way of using information from tissue segmentation (different tissue classes have greatly different relaxation). In our own toolbox Osprey (which can serve as an LCModel wrapper) we circumvent all this by setting atth2o and attmet to 1 and wconc to 65.5M and then doing all relaxation correction outside of LCModel, which works fairly well and is especially advantageous because we can apply different T2 corrections to different metabolites (we know they differ).
EDIT: Even at short-TE, the LCModel defaults may not hold up well since T2s will be shorter at 7T than at 1.5 or 3T. If you have literature values for mice (I don’t), you can just plug them in to get an idea of what atth2o and attmet should be.