vbstatmap
From VoxBoWiki
vbstatmap is a command-line tool that makes it easy to create statistical maps from a VoxBo GLM directory.
To produce a given statistical map, vbstatmap needs to know the following:
- the path of the glm directory
- a filename for the stat cube to be produced
- the desired contrast
Optionally, for t-maps the user may use variance smoothing. Note that the resulting statistical values are not really t-values, and cannot be tested against the t-distribution. They are generally used with a Permutation test.
The best way to specify the desired contrast is using a Contrast file, because that way you can refer to a contrast by name (and contrary to the help text, it doesn't need to be in quotes unless it has spaces in it). You can also specify a contrast vector (weights for each of the covariates of interest) on the command line, but we don't recommend it.
Here's the help text:
VoxBo vbstatmap (v1.8.5pre4/Aug 22 2008)
summary: creates a stat map from a GLM directory
usage:
vbstatmap <glm dir> [flags]
flags:
-s <scale> scale, default creates a t-map
-q <num> q value for an FDR test
-c <contrast> contrast name or vector
-o <file> output file
-p <x> <y> <z> FWHM in voxels for pseudo-t variance smoothing
-m <path/file> relative path and file name to mask cube.
example:
vbstatmap myglm_directory -o out.cub -c "1 0" -s 0 0 0
scales:
t - t value map
i - intercept term percent change map
beta or rb - raw beta values map
f - F value map
tp tp/1 tp/2 - p map of t values
fp - p map for F values
tz tz/1 tz/2 - Z map of t values
fz - Z map for F values
notes:
if you use -c to specify a conrast vector, include only covariates
of interest, and enclose the vector in quotes, e.g.: -c "1 0 -1"
use the -p option to smooth the variance map for a pseudo-t map
note that VoxBo defaults to one tailed tests, and that
/1 and /2 force one tailed and two tailed for p and z maps
