Does anyone know how water is fitted in LCModel if DOWS=T?
For context, we fit our Bruker FID-MRSI data with LCModel and we’re having trouble with absolute concentrations – our relative concentrations are reasonable but when using water as the internal reference, we get overestimations for all metabolites. We have an acquisition delay in our sequence that introduces a first-order phase evolution and it’s accounted for in our basis set. So we’re wondering how water is fitted in LCModel when set as the internal reference, and if the first-order phase in our data influences the accuracy of the fit and quantification. If anyone has any insight on this, we would greatly appreciate it.
LCModel calculates a simple integral over the water signal that is stored under Area of unsuppressed water peak in the .print output files. You should probably validate those outputs by correlating them against manual integration just to make sure this works well for heavily phased data.
Even if the water integral is computed correctly, there are various other factors contributing to the ‘concentration estimate’ that LCModel returns.
It also calculates an integral under a reference singlet in the basis set (WSMET) that is then compared to the water integral (and corrected by the number of protons contributing to this resonance). That is, by default, the Cr methylene signal around 3 ppm (contributing 3 protons). If that signal is heavily phased in your basis set, I can imagine this area calculation being thrown off. Again, probably worth validating.
LCModel makes assumptions during the concentration estimation, namely, it assumes a pure white-matter voxel in terms of water content in the WCONC variable (by default 35.88 M, i.e., ~65% MR-visible water content), and also assumes a fixed relaxation correction factor in the atth2o variable (default value 0.7, i.e., a TE/T2 ratio of approximately 30/80). This assumption is not valid if you go to super-short TE, so you’ll probably want to choose this variable closer to 1 - LCModel does not calculate atth2o (or its metabolite equivalent attmet) from the TE automatically (it also does not know about the T2s of water or metabolites), so it needs to be entered in the control file. What I’d generally recommend anyway is to set the atth2o and attmet factors to 1, set WCONC to pure water (55500 mM) and do your relaxation correction (if necessary) outside of LCModel.