The quantitative parameter estimation code requires two extra header information, tSequenceVariant (Hash of the Pulseq sequence) and tSequenceString (short string that identifies sequence type) for ensuring the proper file naming of the results. These are added to the parameter map files for VD&VE line and NX line under the ParameterMaps/ directory.
In the case of the online reconstruction, these maps need to be copied to the scanner, and pointed out at the XML file at Ice Fire xml config.
In the case of the offline reconstruction, raw data needs to be converted to the MRD format. dat2mrdv2.sh script can do that job.
Gadgetron config files are stored under configs/ directory. Some configs have parameters to send to MATLAB function that effects the mapping algorithm, under the <external><configuration \> tag.
In the case of offline reconstruction, recon.sh script takes the config file as an input.
In the case of online reconstruciton, Ice Fire xml config points out to the desired config file. The xml files need to be copied to Gadgetron installation's config directory, potentially under the conda environment in which the Gadgetron server is called.
For offline reconstruction, use recon.sh.
MRD data directory hierarchy:
|-MRD_ROOT\
|-h5\
|-measXXX.h5
.....
|-noise\
|-noise_measXXX.h5
......
|-measXXXX.dat [optional]
Input format is ./recon.sh /path/to/MRD_ROOT/measXXX[.dat] configs/PulseqXDCartesianXXX.xml.
-
StimFitalgorithm requires information about excitation and refocusing pulses and gradients. It is read fromt2_mese_rfinfo.mat. Pulseq MCSE sequence generation script creates this file. -
Pulseq messes up the headers for 2D sequences since by default scanner thinks the sequence is 3D. This causes kz dimension to be size 2, rather than 1.
passthrough.pypython scripts corrects the data as much as it can. This is why it is only called for 2D sequences.