Hello all,
I am currently looking into preforming tissue correction and have been looking into using Gannet. However, my data was pre-processed (FID-A) and fit (LCModel) not in Gannet and I was wondering if I could import that information into Gannet and then preform only GannetCoRegister onwards. My data was collected on a 3T Siemens scanner using the CMRR MEGA-sLASER (eja) sequence and so from what I understand Gannet isn’t compatible with that yet for fitting (correct?).
In short I am asking: if I have the LCModel output already is there a way to “skip” over importing the raw data and instead import the fitted data + T1w scans to do tissue correction? I am looking for any guidance!
Thank you in advance!!
Hi @mona.dlikan,
Gannet can indeed process MEGA-sLASER data. Some tweaking to the code may be needed to ensure they are loaded properly as we’ve not handle many MEGA-sLASER datasets to date, but that’d be simple enough to do.
What format are your raw data in? Hacking Gannet to import preprocessed data would be a little tedious, so it would be much easier (and methodologically “cleaner”) to preprocess and then fit the data in Gannet.
Usage instructions can be found on our documentation website.
Mark
Oh amazing, thank you so much! I assume the main changes would live in the initialization script – so for things like seqorig if none of the expected options are compatible where would I add information about the sequence I have used specifically?
My raw data is .dat, which I see should be compatible.
Assuming you are using the latest version of Gannet, seqorig is not applicable to Siemens data. Gannet should be able to recognize that you used the CMRR MEGA-sLASER sequence.
Other than that, yes, make sure to set any of the other parameters in GannetPreInitialise.m appropriately for your data.
1 Like
Hello,
In trying to avoid having to re-process and re-fit all my data I was looking through and found the script “GannetMask_SiemensTWIX.m” – could I not load my raw data and a structural image into this & then use the generated mask to preform segmentation using another software like FSL-FAST? Or even then from there preform just the segmentation in Gannet? Alternatively, is there another way to just do the registration and segmentation in Gannet (or another software) and then I could apply those results to my already fitted data manually (in theory).
Otherwise, is there a script hidden somewhere which allows me to read the angulation and position of my voxel from the .dat files – with that I can do my own voxel mask creation.
Hi @mona.dlikan,
You can use CoRegStandAlone.m to co-register voxels and segment anatomical images without running the MRS data analysis. The usage is:
MRS_struct = CoRegStandAlone({'mrs_file.dat'}, {'T1_img.nii'});
You can also batch-process data by including multiple filenames in the cell arrays.
Note, however, that by doing the co-registration and segmentation separately from the main pipeline the tissue fractions will be saved in a separate output structure, so you’d need to intergrate them into your previous Gannet output structure if you wanted to run GannetQuantify.m on your MRS data.
Mark
Thank you so much! That’s great news(:
For the MRS file would I be importing the edit off or on acquisition (or maybe the WSCal)? Maybe that is a question for CMRR about which of the files outputted from their sequence holds the relevant voxel information.
Any MRS file would work as they all will contain the relevant voxel information in the header (assuming you didn’t move the voxel in between acquisitions).
1 Like