Basis set request for 3T MEGA-sLASER GABA quantification

Analysis / Modeling Basis Sets
1

Hi all,

I am working on MRS analysis to quantify GABA. My data has been collected on a 3T SIEMENS MAGNETOM Prisma using the CMRR Marjanska/Auerbac MEGA-sLASER sequence with TE = 68ms. I have built a pre-processing pipeline from FID-A and am ready to move onto LCModel to fit my data but am struggling to find a basis set. I am wondering if anyone has any basis sets or knows of where I could find some out of the box ready to use for my project.

Any guidance would be greatly appreciated! Thank you in advance(:

2

I have used MRS cloud, https://braingps.mricloud.org/home

where you can simulate your basis set for desired vendor, TE and sequence, and it worked for me

3

Thanks, I have also been exploring these simulation softwares along with my search for an out of the box solution! I just need to iron out some details with the physicists here to confirm the details of acquisition.

Might be a naive question but what is the difference between ā€œSiemensā€ and ā€œUniversal Siemensā€ when selecting the vendor? I perused through the code and canā€™t figure it out exactly.

4

@mona.dlikan, the Universal Siemens option helps simulation of basis set using the standardized rf pulses and timings (doi: 10.1016/j.neuroimage.2019.01.056) at Siemens B0. Select ā€˜Universal Siemensā€™ only if youā€™re using the universal sequence. Otherwise, select ā€˜Siemensā€™.

1 Like
5

Thank you!

One last question, I donā€™t see the option to include macromolecules in MRSCloud simulation. GABA is my main metabolite of interest so I want to be sure to include the correct MMs which overlap with GABA (I assume?). Does MRS cloud automatically include these (aka GABA+ is assumed when GABA is included) or should I simulate the .jsonā€™s of my macromolecules of interest and then use the simulated .jsonā€™s from MRSCloud to build my own .basis?

I also have an issue of some lipid contamination (in the macromolecular frequency range) in my data which I was able to eliminate mostly through my pre-processing and development of my difference spectra. Does this also need to be considered in my basis set?

6

MRSCloud currently does not support MEGA-sLASER with the CMRR pulses or timings. Iā€™m currently trying to figure out how to get this sequence in there based on @wclarkeā€™s excellent detective work, but Iā€™m still not sure Iā€™ve fully gotten it right, and Willā€™s repo does not include the editing pulse used at 3T.

For now, until we have figured out how to accurately simulate this, I would really recommend reaching out to the CMRR folks for LCModel basis sets for this particular setup.

As to the macromolecules, they will not be included in the LCModel basis sets created by MRSCloud anyway. You will need to parametrize those in the LCModel control file. We have previously done some work to figure out a good parametrization. The way we tell LCModel to use it is in osprey/fit/code/osp_fitInitialise.m at develop Ā· schorschinho/osprey, lines 554 and following:

nsimul = 2
chsimu(1) = 'MM09 @ 0.915 +- .02 FWHM= .085 < .1 +- .35 AMP= 3.'
chsimu(2) = 'MM30 @ 3.0 +- .02 FWHM= .085 <  0.12 +- .35 AMP= 2.'
chrato(1) = 'MM30/MM09 = 0.66 +- .2''
ppmgap(1,1) = 1.95
ppmgap(2,1) = 1.2
7

That puts a wrench in things but good to know! Thank you for your guidance.

Do you have any other recommendations for basis set simulation or where to look? I reached out to CMRR and am waiting for a response, but want to have backup options as well just incase!

8

@admin

I think Iā€™ve sorted it now. I also think I had the pulses / cases the wrong way around. Will double check and upload in the next few days.

1 Like
9

@admin and @mona.dlikan

The updated sequence description is now available at sequences/eja_svs_mslaser Ā· master Ā· William Clarke / WIN MRS Basis Sets Ā· GitLab

It does seem to need a fairly high number of spatial simulation points to work well. So @admin , if you do ever get a chance to check / run through your own simulator Iā€™d love to know that Iā€™ve got it right.

1 Like