MEGA-PRESS acquisition parameters for later analysis in Osprey

Dear Osprey/MRS experts,

I am currently designing/piloting an MRS study using the Johns Hopkins MEGA-PRESS sequence (metabolite of interest is GABA). I plan to use Osprey to analyse my data but am running into issues during analysis that I suspect are due to my acquisition parameters.

Oeltzschner et al.'s 2020 paper (doi: 10.1016/j.jneumeth.2020.108827), as well as the Osprey jobfile and manual suggest that three files should (or can) be used in the analysis: the metabolite file (in my case using a MEGA-PRESS acquisition); lineshape reference data (acquired with the same sequence as the metabolite data, but without water suppression, and used to perform eddy-current correction); and short-TE reference data.

We are using a Siemens Prisma 3T scanner. When we attempted to acquire the lineshape reference data (water unsuppressed data), our scanner automatically changed the delta value to one that did not match the original MEGA-PRESS acquisition parameters. Should I leave this or adjust to match the original (water suppressed) sequence?

Secondly, I would like to know the recommended number of volumes for the lineshape reference and short TE reference scans. In my initial attempt, I tried collecting a single volume, which I was told would be sufficient but I suspect that this is contributing to the errors I am getting during analysis in Osprey. Should I perhaps be collecting 8 or even 16 volumes for these data?

Lastly, I wondered if anyone who has successfully acquired and analysed MEGA-PRESS data (using the TWIX files) in Osprey would mind sharing their sequence information please? This would make it much easier to see where we have gone wrong.

I attach my Osprey jobfile but as mentioned, I think the issues stem from the data rather than the program.osprey_jobfile.docx (18.6 KB)

On loading the data into Osprey, I get the error message:

Error using cat
Dimensions of arrays being concatenated are not consistent.

Error in io_loadspec_twix (line 401)
    fids = cat(4,fids_A,fids_B,fids_C,fids_D);

Error in osp_LoadTwix (line 71)
            raw_ref                     = io_loadspec_twix(MRSCont.files_ref{kk});

Error in OspreyLoad (line 60)
                [MRSCont] = osp_LoadTwix(MRSCont);

Error in osp_onLoad (line 33)
    MRSCont = OspreyLoad(MRSCont);
Error while evaluating UIControl Callback.

The code seems to be treating the data as HERMES not MEGA-PRESS judging by the placement of the error (line 401). If I comment out this code, I get an error not much further downstream. If I attempt to run the jobfile using no input for the files_ref, I can load the data, but get an error later in processing. Happy to share these details but they may just be due to the data I’m trying to load in.

Many thanks for your help!

Carolyn McNabb (MRS novice)

1 Like

Hi Carolyn,

We’re happy to help you make this work, and it’s great that you’ve decided to collect same-TE and short-TE data. Most people still don’t bother getting two water references when the added acquisition time is really only a few seconds!

Now, for your questions:

  1. You absolutely want the delta frequency to be different between your water-suppressed and your water-unsuppressed data. The reason is the chemical shift displacement effect, which is nicely explained in the respective section under this link. You want to make every effort to have the spatial origin of your water signal to be as co-localized as possible with the spatial origin of your metabolite of interest. The delta frequency determines the transmitter frequency of the slice-selective radiofrequency pulses, and (on the Siemens systems) is determined relative to water. So, if you set the delta frequency to 0.0 ppm, the voxel you prescribe on the screen will be the one where your water signal comes from (while the other metabolites are slightly shifted, according to the CSDE); if you set it to -1.7 ppm, the voxel you prescribe will be the one where the 3.0 ppm (= 4.7 ppm for water - 1.7 ppm) signals come from. (The considerable CSDE with traditional PRESS is also one major reason why adiabatic localization like (semi-)LASER with high-bandwidth pulses is becoming more popular.)

1b. If you would like to have your acquisition parameters double-checked, feel free to post the exported PDF of your protocol here (or via e-mail). There are always some tiny settings that you may overlook.

  1. I think that you should get away with acquiring only 1 water transient, and I’m pretty sure the Osprey error here is caused by the assumption that the water reference has at least as many averages as there are sub-experiments in the acquisition (4 for HERMES, 2 for MEGA).
    (NB, by volume MRS people tend to understand the measurement volume rather than repetitions of the same sequence, which are usually referred to as transients or numbers of excitation)

  2. Your job file looks fine, really, we’ve just not tested this on single-transient water reference data and need to add a couple if statements, I guess. Would you mind passing this dataset on to us (after de-identifying using TwixDeidentify)?

Happy to help further.


This is so helpful. Thank you, Georg!

I have created a new folder in the OneDrive directory I shared with you previously. It includes the TWIX files and anatomical scan. Luckily we acquired the water unsuppressed data using both delta values - I’ve only included the correct one in the folder.

Thank you so much for offering to look over the acquisition parameters too! I will hopefully have the PDF of the scanning protocol tomorrow but next week at the latest. Also, thank you for explaining everything so clearly. It really helps.

Best wishes,

1 Like