GLM and related operations

glm_4dfp

multivariate voxelwise regression/correlation

Usage: glm_4dfp <format|fmtfile> <profile> <4dfp|conc input>

Examples:

glm_4dfp "4x124+" doubletask.txt b1_rmsp_dbnd_xr3d_norm

Options

-Z	supress automatic removal of mean from input regressors
-C<str>	read partial beta coefficients from specified 4dfp image (default compute)
-o[str]	save partial beta images with specified trailer (default = “coeff”)
-R	compute partial beta images as percent modulation
-b[str]	save total beta images with specified trailer (default = “tbeta”)
-p[str]	save partial corr images with specified trailer (default = “pcorr”)
-t[str]	save total corr images with specified trailer (default = “tcorr”)
-r[str]	save residual timeseries with specified trailer (default = “resid”)
-@<b|l>	output big or little endian (default input endian)

N.B.: conc files must have extension “conc”
N.B.: <profile> lists temporal profiles (ASCII npts x ncol; ‘#’ introduces comments)
N.B.: <profile> line limits are 81920 chars and 8192 fields
N.B.: absent -C, options -o and -r require design matrix inversion; dimension limit 256

actmapf_4dfp

voxelwise evaluate timeseries inner product against reference waveform

Usage: actmapf_4dfp <format|fmtfile> <4dfp|conc input>

Examples:

actmapf_4dfp -zu "3x3(11+4x15-)" b1_rmsp_dbnd_xr3d_norm
actmapf_4dfp -aanatomy -c10 -u "+" ball_dbnd_xr3d.conc
actmapf_4dfp -zu "4x124+" b1_rmsp_dbnd_xr3d -wweights.txt

Options

-a<str>	specify 4dfp output root trailer (default = “actmap”)
-c<flt>	scale output by specified factor
-u	scale weights to unit variance
-z	adjust weights to zero sum
-R	compute relative modulation (default absolute)
-w<weight file>	read (text) weights from specified filename
-@<b|l>	output big or little endian (default input endian)

N.B.: conc files must have extension “conc”
N.B.: when using weight files ‘x’ frames in format are not counted
N.B.: relative modulation images are zeroed where mean intensity < 0.5*whole_image_mode

GC_4dfp

Granger causality mapping Usage: GC_4dfp <format> <4dfp|conc input> <order>

Examples:

GC_4dfp "4(4x190+)" VB20579_rmsp_faln_dbnd_xr3d_atl.conc 2

Options

-w<str>	specify timecourse profile file (one or more columns)
-i<int>	use only specified column (counting from 1) of timecourse profile
-a<str>	append specifed string to map output
-g	write lagged covariance (gamma) 4dfp stack
-D	write difference of directed influences (Fx->y - Fy->x) map
-Z	write Geweke “N(0,2)” measure (difference of square roots) map
-F	write Fx,y, Fx->y, Fy->x, Fx.y map stack
-@<b|l>	output big or little endian (default input endian)

N.B.: conc files must have extension “conc”
N.B.: effective frame count is determined by <format>
N.B.: ‘x’ frames in format are not counted

GC_dat

Granger causality on ASCII column data

Usage: GC_dat <format> <input_datafile> <order>

Examples:

GC_dat "4x106+" ROI_timeseries.dat 2

Options

-d	debug mode
-v	verbose mode
-u	normalize all input timeseries to unit variance
-x<int>	specify dimensionality of x process (default = 1)
-m	create text listing of AR model
-w	write residual after full AR modeling
-P	format residual output suitable for plotting (xyy)

covariance

covariance, correlation, coherence, etc. on ASCII column data

Usage: covariance <format|fmtfile> <profile>

Examples:

covariance "4x124+" doubletask.txt

Options

-q	quiet mode
-t	optionally remove trend (ramp) from input timeseries
-u	optionally normalize all input timeseries to unit variance
-o	output lagged CCV dat files (CCR with -u)
-a	output lagged ACV dat file (ACR with -u)
-r	output Bartlett smoothed cross spectra (spectral density with -u)
-p	output Bartlett smoothed auto spectra (spectral density with -u)
-e	compute eigenvectors of lag 0 CCV
-L	read ROI labels from <profile> (default ignore ‘#’ commented lines)
-T<int>	additionally smooth spectra with Tukey window of specified width (in frames)
-d<flt>	specify frame TR in sec for Fourier analysis (default = 1.0000)
-m<int>	specify CCV function maxumum lag in frames (default = 32)
-D<flt>	SVD lag 0 CCV and output new profile with cndnum < specified value (implies -e)
-g<str>	regress timeseries in named file out of <profile>

N.B.: all input timeseries are made zero mean as a first step
N.B.: region names can be specified on the first line of <profile> with ‘#’ in the first column