This is an old revision of the document!
Currently this assumes that you have recorded your physiological responses through Cigal. Also, these types of corrections should be done before any type of slice-time correction or pre-processing.
There is currently a tool for running AFNI-based correction of functional image data using physiological data. In the future it may support FSL-based correction.
Usage: physio_run.py /path/to/run4.bxh /path/to/pdigm5_12345_4_2 OUTPUTPREFIX [ -f FORMAT ] Program to produce formatted physiological data: Data will be corrected and recreated from the pdigm file based on information from run data BXH and PDIGM required OUTPUTPREFIX is the prefix for the physio-corrected output FORMAT: what pipeline to use, only supports 'afni' for now (default) Options: -h, --help show this help message and exit -f string, --format=string output format ( afni )
physio_create.py under the hood to write physiological data in the format AFNI requires (see below for how to run
physio_create.py yourself to create data usable by FSL), and then runs various AFNI tools to actually create the physio-corrected image data.
The outputs will be OUTPUTPREFIX.bxh, OUTPUTPREFIX.nii.gz, OUTPUTPREFIX_cardiac.txt and OUTPUTPREFIX_respiration.txt. OUTPUTPREFIX.bxh can be used as the input for
The underlying tool used by
physio_run.py to create the regressors is
physio_create.py. It takes an input image along with your cigal pdigm file to create various types of corrected text files to be used for physio corrections in a number of different packages.
This tool uses various fields from the bxh and pdigm file to output “corrected” (physiological) data/regressors. “Corrected” here means accounting for potential time-locking issues, different TR in the cigal recording verses the actual data acquisition, ddqs, etc.
The python tool will call cigal functions directly to do the corrections, then output new text data in the format requested.
physio_create.py --bxh /path/to/run4.bxh --pdigm /path/to/pdigm5_12345_4_2 -f fsl --outpath /here/ --hz 100 Program to produce formatted physiological data: Data will be corrected and recreated from the pdigm file based on information from run data BXH and PDIGM required OUTPATH defaults to PWD HZ sampling rate FORMAT: defaults to fsl - fsl = cardio,resp,TR pulse ( 3 column file ) - npm = time,cardio,resp ( 3 column file ) - afni = cardiac.txt, respiration.txt, script.m Options: -h, --help show this help message and exit -b FILE, --bxh=FILE bxh file for run -p FILE, --pdigm=FILE pdigm file for run -f string, --format=string output format ( fsl,npm,afni ) --hz sampling rate is hz ( 100 ) -o PATH, --outpath=PATH location to store output files
This section details the steps that the tool
physio_run.py (described above) does for you automatically.
Create the AFNI formated physiological text data.
physio_create.py -b Data/Func/20111025_12345/run004.bxh -p Data/Behav/12345/pdigm5_12345_4_2 -f afni --hz 100
This would create my “cardiac.txt” and “respiration.txt”, and a MATLAB script script.m that will create the regressors, and has the following contents:
addpath /usr/local/packages/MATLAB/afni/ Opt.Respfile = '/path/to/respiration.txt' Opt.Cardfile = '/path/to/cardiac.txt' Opt.VolTR = 1.5 Opt.Nslices = 34 Opt.PhysFS = 100 Opt.SliceOrder = 'alt+z' %please choose the correct order if not running interleaved: **'alt+z'**/'alt-z'/'seq+z'/'seq-z' RetroTS(Opt)
The output of runnning the MATLAB script will be “oba.slibase.1D”, which is a text file containing regressors to remove from your data on a slice-by-slice basis.
Convert your data to 4D nifti if you haven't already done so:
bxh2analyze --niigz -s input.bxh run004
Create the afni script which you will use to actually run the 3dretroicor functionality:
afni_proc.py -subj_id 12345 -dsets run004.nii.gz -blocks despike -do_block ricor -tcat_remove_first_trs 0 -ricor_regs *.slibase.1D -ricor_regs_nfirst 0 -ricor_regre ss_method 'per-run'
The result of the above command would be a tcsh script, which would run the despiking and 3dretroicor correction only. Run it:
tcsh -xef proc.12345 |& tee output.proc.12345
There will be a resulting folder with all the afni data inside. You can convert the afni HEAD/BRIK to a nifti file, then proceed to whatever you're doing afterwards:
#convert afni to nii, create bxh bxh2analyze --niigz -s 12345.results/pb02.12345.r01.ricor+orig.BRIK physiocorrected
Run the resting state pipeline:
resting_pipeline.py -f physiocorrected.bxh -s 0,1,2,3,4,5,6,7 -p func --sliceorder odd
We've gotten the FSL beta functions for physiological correction, which we'll test.
They are based on the same algorithms used in afni with retroicor. Once things are functioning, they'll be installed on the cluster.
More information can be seen here:
The beta options will allow you to plug in the resulting regressors directly into Feat. After pre-processing you can use the res4d.nii.gz files for further analysis ( filtered_func_data would not have the physio regressor removed ).