Hi all, I’m transitioning from Gannet to Osprey and have several questions.

I want to use the alpha tissue correction. Looking at MRSCont.quantify.metab.TissCorrWaterScaled - this is a structure of size 1 x nFiles x 2, and each cell contains 30x1 vector, representing the 30 possible metabs, which is what I would expect. But in MRSCont.quantify.metab.AlphaCorrWaterScaled, each cell contains nFiles values - why is that? I’m not sure how to use this.

Relatedly, what is a good way to apply the alpha correction to Glx? It’s not included in the .csv files, and I can’t read it from the MRSCont (as detailed in 1).

Is there some index of “goodness of fit” that could be used for individual fits quality assessment?

The MRSCont.quantify.metab.AlphaCorrWaterScaled is for Osprey’s internal data handling. If you want to see the correct number you should look into the tables (MRSCont.quantify.tables.metab.AlphaCorrWaterScaled.Voxel_1). The dimensions will be nsubjects x nmets with nmets being the number of alpha corrected metabolites.

I have just added alpha correction for Glx to the most recent version on the develop branch. It has the same ratio as GAAB assuming a ratio of 1/2 for WM/GM. You can give that a try with your data after downloading. I have also fixed a scaling issue for the 1to1GABA model just in case your quantification numbers change.

Osprey only has a relative residual number to assess the fit quality for now. You can find this in the QA tsv file in the output folder. relResA and relResdiff1 are the normalized residuals. Unfortunately, there are currently no other uncertainty metrics available.

Let me know fi you need more help during the transition from Gannet to Osprey.

I downloaded the current version on the develop branch, and I’m still not getting Alpha correction for Glx, only GABA (both in MRSCont.quantify.tables and in quanfify/diff1_AlphaCorrWaterScaled_Voxel_1_Basis_1.tsv).
Any suggestions for what could have gone wrong?