Afni3dDeconvolve
Afni3dDeconvolve is a wrapper for the AFNI (RW Cox, NIH) 3dDeconvolve application.
3dDeconvolve -- Multiple linear regression of a 3D+time series, including
linear deconvolution to find an ideal response vector for each voxel.
AFNI 3dDeconvolve was written by B. Douglas Ward, Medical College of
Wisconsin.
For more information about the AFNI package, see the AFNI
Homepage
To see the AFNI postscript documentation please go here.
To read the help documentation generated from the command line go here.
To learn about afni data types follow this link.
Invocation
java Afni3dDeconvolve
This starts up the java interpreter and runs the Afni3dDeconvolve application.
You need to have set up your environment
for java in order for this to work.
For information on how to invoke the command line program, 3dDeconvolve,
please go here.
Afni3dDeconvolve window
Information to be entered
-
Input file
AFNI 3d+time dataset.
-
Stimulus files
ASCII files of covariates. One covariate per column.
-
Output bucket
AFNI output bucket name. The output bucket will be placed in the current
working directory.
-
Select covariates...
Bring up the window to select covariates and set lag, irf, std parameters.
-
Editor pane
-
Filename: Filename where the covariate comes from.
-
Stim #: Number of the covariate in the set to be included.
-
Include: flag to include this covariate in the regression or not.
-
Col #: Column number of this covariate in the stimulus file.
-
Label: Label to call this covariate in the output bucket. (-stim_label
flag)
-
Min Lag, Max Lag: Set the number of lag times at which to include this cov.
(-stim_minlag and -stim_maxlag flags)
-
Save irf: flag to save the IRF (lag betas) for this covariate. (-iresp flag)
-
Irf prefix: The prefix for the 3d+time dataset that will be written with the IRF.
-
Save std: flag to save standard deviations for IRF. (-sresp flag)
-
Std prefix: The prefix for the 3d+time dataset that will be written with the standard deviations for the IRF.
-
Nptr: specify that there are n (default to 1) timepoints per TR for this covariate. (-stim_nptr flag).
Note that to use this option your dataset must have had all slice timepoints within
each volume aligned to 0 time offset. Also, the input stimulus file cannot be
"ragged", all columns must have the same number of rows. Shorter columns can be
padded with 0s to get equal lengths.
-
Buttons
- Reload files: load/reload the covariates in the files entered in the main window Stimulus file textbox into the editor pane.
- Include All: shortcut to include all covariates in regression.
- Exclude All: shortcut to exclude all covariates from regression.
- Edit: bring up editor box to set parameters for highlighted covariate.
- Set Defaults: bring up editor box to set parameters for all covariates.
-
Select glt tests...
Editor to select and create general linear tests to be performed.
-
Select glt tests... Name
Name of the glt file. Entries are added to this box when the Add GLT to Regression
button below is clicked. All entries in this box will be included as general
linear tests in the multiple regression.
-
Select glt tests... Label
Label for the glt in the output bucket. Entries are added to this box when the
Add GLT to Regression button below is clicked.
All entries in this box will be included as general linear tests in the
multiple regression.
-
Select glt tests... Clear
Clear the selected entry from the Name/Label boxes and thus remove it from the
multiple regression.
Select an entry by clicking on it in the Name box.
-
Select glt tests... Clear All
Clear all entries from the Name/Label boxes and thus remove all general
linear tests from the multiple regression.
-
Select glt tests... GLT file
Select a file containg a general linear test specification.
-
Select glt tests... GLT label
Enter a label for this GLT that will appear in the 3dDeconvolve output bucket.
-
Select glt tests... Create GLT with n rows
Open a spreadsheet dialog to create a general linear test file.
You must specify how many rows (constraints) will be in your general
linear test before opening the spreadsheet via Create GLT.
-
Select glt tests... Add GLT to regression
Add the GLT specified in the file/label fields to the list (upper half
of the GUI) of general linear tests to be included.
-
Options...
Please see the AFNI postscript documentation
for a description of these commandline options.
Note that commandline options not present in this dialog box are entered
via the Select covariates... or Select glt tests... dialog boxes.
Special runs
-
Options... Check collinearity only
-nodata commandline flag
calculate inverse of design matrix only, to check for collinearity,
do not compute multiple regression.
-
Options... Run 1 voxel only
-input1D dname commandline flag
compute the regression only on the vector in dname, AFNI 1D
ascii file.
Input options
-
Options... Use mask file
-mask mname commandline flag
compute regression only for voxels in the maskfile.
-
Options... Use censor file
-censor cname commandline flag
Data timepoints set to 0 in cname are excluded from the analysis.
See Section 1.4.7.
-
Options... Use concat file
-concat rname commandline flag
rname is an AFNI 1D file containing start points for each run.
Compute multiple regression taking run timing into account. See Section
1.4.6.
-
Options... Set image range
use -nfirst -nlast commandline flags. See below.
-
Options... first
-nfirst fnum commandline flag
0 based indexing. fnum is the first timepoint to be used in
computing multiple regression.
- Brick...
-
Options... last
-nlast lnum commandline flag
0 based indexing. lnum is the last timepoint to be used in computing
multiple regression.
- Brick...
-
Options... Detrend polynomial order
-polort pnum commandline flag
Degree of the polynomial in the baseline model. Off means do not add
a trend covariate to the model, 1 for a linear trend.
-
Options... RMS min
-rmsmin r commandline flag
compute full regression model only for voxels whose time series when
fitted with the baseline model, has error rms < r. Use a mask
file as alternate method to exclude voxels.
Output file options
-
Options... Save F stats
-fout commandline flag
flag to output partial and full model F-statistics
-
Options... Save T stats
-tout commandline flag
flag to output the t-statistic for each regression parameter
-
Options... Save R-sq stats
-rout commandline flag
flag to output partial and full model R squared statistic
-
Options... Save MSE map
-vout commandline flag
flag to output the sample variance (MSE) map
-
Options... full first
-full_first commandline flag
flag to specify that the full model statistics will appear first in
the bucket dataset output.
-
Options... No coeff stats
-nocout commandline flag
Flag to suppress output of fit coefficients (and associated statistics).
-
Options... No B0 stats
-nobout commandline flag
Flag to suppress output of baseline coefficients (and associated statistics).
-
Options... tshift irf data
-tshift commandline flag
Use cubic spline interpolation to time shift the estimated impulse
response function, in order to correct for differences in slice acquisition
times. Note that this effects only the 3d+time output dataset generated
by the -iresp option.
-
Options... Save fitted data
-fitts fprefix commandline flag
fprefix = prefix of 3d+time output dataset which will contain the (full
model) time series fit to the input data.
-
Options... Save residual data
-errts eprefix commandline flag
eprefix = prefix of 3d+time output dataset which will contain the residual
error time series from the full model fit to the input data
Screen display options
-
Options... Display F stats over
-fdisp fval commandline flag
Write (to screen) results for those voxels whose F-statistic is > fval
-
Options... Display X matrix
-xout commandline flag
flag to write X and inv(X'X) matrices to screen
-
Options... Display progress every n voxels
-progress n commandline flag
flag to write progress report to screen every n voxels.
-
Number of jobs
-jobs J commandline flag
J = number of jobs (sub-processes) to run the program in parallel.
Use this if you are running on a multi-CPU machine.
The maximum number of jobs that can be specified is 32.
Last updated Thu Oct 11 16:43:37 EDT 2001