3D_SOSP_fMRI

This repository houses the sequence and analysis scripts for the paper Combining the benefits of 3D acquisitions and spiral readouts for VASO fMRI at UHF [https://doi.org/10.1162/imag_a_00308]

Sequence (pulseq):

Intial configuration steps:

1. Clone this repository

   git clone https://github.com/monreal93/3D_SOSP_fMRI.git

2. clone pulseq toolbox into ./tools/pulseq/

   git clone https://github.com/pulseq/pulseq.git

3. clone PNS prediction into ./tools/pns_prediction/

   git clone https://github.com/filip-szczepankiewicz/safe_pns_prediction.git

4. The main file to generate pulseq sequences is ./pulseq/create_pulseq.m

5. Check system limits, if yours is not included, add an extra option for the params.gen.field_strength field and include the system limits in function ./pulseq/functions/prepare_system_limits.m current systems limits:

   • SIEMENS 7T Magnetom Plus with SC72 gradients
   • SIEMENS 9.4T with AC84-II gradients
   • SIEMENS NextGen 7T with impulse gradient

6. If you want to check for PNS, inlcude your gradient .asc files in the folder ./tools/pns_prediction/gradient_files and use the flag params.gen.pns_check=1

7. To generate a sequence run the file ./pulseq/create_pulseq.m , different parameters are to be set in the section "Define parameters":

   • folder_name = "folder_name" -> Folder where the pulseq and trajectory files will be saved
   • seq_name = "seq_name" -> Sequence name (any name you want to use)
   • params.gen.seq = "sequence_type" At the moment the well tested sequences are 1-VASO and 4-BOLD

8. After the script runs a folder ./data/"folder_name" will be created, with the following subfolders:

   • acq ->
   • analysis ->
   • ismrmd ->
   • raw ->
   • recon ->
   • sim ->
   • tmp ->

9. Several files will be written into the folder ./data/"folder_name"/acq , this files will be used to run the sequence at the scanner and to reconstruct the data:

   • "seq_name".seq -> pulseq sequence file
   • "seq_name"_params.mat -> directory with sequence parameters
   • "seq_name"_ks_traj.mat -> nominal trajectory

10. Once you acquire the data, save it in ./data/"folder_name"/raw/twix/

RECONSTRUCTION (MRIReco.jl):

Pre-requisits:

1. ISMRMD
2. Vendor raw data format to ISMRMRD converter
3. VSCode with remote explorer and julia extensions

Initial configuration steps:

1. Get Julia docker image docker pull julia:1.8.5 and create a container

   docker run -t -d -P -v "path_to_folder":/usr/share/sosp_vaso/ --name julia_mri_recon julia:1.8.5
   docker start julia_mri_recon

2. I advise using VS code, follow the instructions to attach to a running docker container from VSCode
3. Install julia extension in container
4. Once attached to contaier, add folder /usr/share/sosp_vaso to workspace and open file ./recon/reconstructions_julia_mrd.jl
5. Open the julia REPL (CTRL+SHIFT+P) Start julia REPL
6. Activate julia enviroment in recon folder:

   using Pkg
   Pkg.activate("./recon/")

7. Add modified version of MRIReco.jl

   ]
   dev https://github.com/monreal93/MRIReco_amm.git
   dev MRIOperators

8. Instantiate (download and install packages):

Pkg.instantiate()

Reconstruct the sample data

1. Download sample raw data from LINK and save it into ./data/
2. Convert vendor spiral and GRE-ME (for sensitivity and off-resonance maps) raw data to ISMRMRD format and save it into ./data/sample/ismrmd/3d/

   siemens_to_ismrmrd -f ./data/sample/raw/twix/sample.dat -o ./data/sample/ismrmd/3d/sample.h5
   siemens_to_ismrmrd -f ./data/sample/raw/twix/b0_sample_fieldmap.dat -o ./data/sample/ismrmd/3d/b0_sample_fieldmap.h5

3. Open VScode and add folder /usr/share/sosp_fmri/ to workspace.
4. Open the julia REPL (CTRL+SHIFT+P) Start julia REPL and type

   cd("/usr/share/sosp_fmri")
   include("./recon/reconstructions_julia_mrd.jl")