[MINC-users] glim_image

Andrew Janke a.janke at gmail.com
Thu Apr 30 18:21:26 EDT 2015 

This looks like a dump from -help which is this:

EXAMPLE - group comparison

   glim_image -max_num_row 350 \
       -t_stat glim_t1-2_groupAB-10mm.mnc t_stat voxel column 2 \
       glim_t1-2_groupAB-10mm

EXAMPLE - pairwise subtraction

   Pfizer_Part2


Command-specific options:
Stats options:
Common options:
 -family:                Specify exponential family for GLM.
 -link:                  Specify link function for GLM.
 -var_func:              Specify variance function for GLM.

 -t_stat:                Defines t-statistics wanted for output
T-statistics are defined as follows: -t_stat <out.mnc> type (one of t_stat,
stdev, corr, beta) stdev_type (one of pool or voxel) intype (one of matrix,
column or file) string (string defining contrast depends on intype)

For example, the following:
    -t_stat out.mnc corr pool column 4
defines an output file that will be a partial correlation testing column 4
of the design matrix

    -t_stat out.mnc beta voxel matrix '0 1.5 0 -2.3'
defines an output file that will have the of 1.5*beta2 - 2.3*beta4
    where the betas are the parameters estimated in the regression.

    -t_stat out.mnc stdev pool file contrast1.txt
defines an output file that will be the stdev of the t_stat defined in the
textfile contrast1.txt with a pooled stdev. This doesn't make sense because
we would not use a pooled stdev to look at the stdev as it varies over the
image, so glim_image spits this out as a warning.

 -f_stat:                Defines F-statistics wanted for output
F_statistics are defined in the same way as t_statistics except there is no
type, this is because the variance of an F-stat is a matrix, as is the
numerator. For this reason, column is also not an intype for F_statistics
because a matrix with more than two rows is needed for an F_stat.

For example:
    -f_stat out.mnc pool matrix '0 1 2 0 ; 1 2 3 0'
defines an output file that is an F-stat testing whether
    beta2 + 2*beta3 = 0 and beta1 + beta2 + beta3 = 0
simultaneously. This comes into play for anova designs when there are
factors
with multiple levels.

 -est_scale:             Use estimated scale (for binomial and poisson
families)
 -scale_par:             Set scale parameter, (default: estimate it)
                Default value: -1
 -max_iter:              Maximum iterations for Fisher scoring.
                Default value: 10
 -tolerance:             Tolerance for convergence of fitted parameters.
                Default value: 0.0001
 -eps_sing:              Epsilon added to repsonse values that would cause
singularities.
                Default value: 0.0001
 -maxVoxelSdBasedValue:  Maximum absolute value of t_statistic image.
                Default value: 99
 -avg:                   File name for means (missing values not skipped.
Needs sd_type and out_type.
Avg is used for comparison of means at the same time as fitting a design
 file, the output types are t_stat, beta, corr and stdev; stdev types are
voxel or pool.
 -deviance:              Output an image of deviances (default: FALSE)
 -avg_deviance:          Output image of deviances for avg.
 -scale:                 Output image of scale estimate (default: FALSE)
 -errfile:               Output convergence error codes.
 -use_mu:                Use previous fitted values instead of response for
initial values.
 -pool_sd:               Mask file for region over which to pool s.d.
 -fwhm:                  Calculate FWHM estimate to be used for assessing
significance.
 -fwhm_gauss:            Output fwhm image for gaussian fit.
 -fwhm_simple:           Output non-weight corrected FWHM estimate for
non-Gaussian data.
 -fwhm_avg:              Output fwhm image for comparison of means fit.
 -search:                Mask file for search region.
 -fit_all:               Fit GLM at all voxels with no missing values, not
just search region.
 -fit_missing:           Fit GLM at all voxels, even with missing values.
Must provide name for
degrees of freedom volume.
 -max_num_row:           Maximum number of rows for design_file. Should be
increased for large files.
                Default value: 300
 -max_num_col:           Maximum number of rows for design_file. Should be
increased for large files.
                Default value: 100
 -scylla:                Flag to see if computer is at neuro or math dept.
General options:
 -path:                  Path for input files and design_matrix file.
 -clobber:               Overwrite existing file.
 -noclobber:             Don't overwrite existing file (default).
 -verbose:               Print out log messages (default).
 -quiet:                 Do not print out log messages.
 -copy_header:           Copy all of the header from the first file.
 -nocopy_header:         Do not copy all of the header from the first file.
 -filetype:              Use data type of first file (default).
 -byte:                  Write out byte data.
 -short:                 Write out short integer data.
 -long:                  Write out long integer data.
 -float:                 Write out single-precision floating-point data.
 -double:                Write out double-precision floating-point data.
 -signed:                Write signed integer data.
 -unsigned:              Write unsigned integer data (default if type
specified).
 -range:                 Valid range for output data.
                Default value: 0 0
 -max_buffer_size_in_kb: Specify the maximum size of the internal buffers
(in kbytes).
                Default value: 10240
 -check_dimensions:      Check that files have matching dimensions
(default).
 -nocheck_dimensions:    Do not check that files have matching dimensions.

Generic options for all commands:
 -help:                  Print summary of command-line options and abort

Usage: glim_image [options] <des.file>
       glim_image -help

Your command line was (after parsing):

      glim_image -help

Use '-' as des.file to read from stdin. Last line should end with 'END'.


I am sure I cached jon's help page somewhere but can't find it right now.
If anyone can find it please point me to it (even in archive.org) and I'll
put it up on the MINC wikibooks pages.


a


On 1 May 2015 at 08:18, Vladimir S. FONOV <vladimir.fonov at gmail.com> wrote:

> Hello,
>
>
> there is a little bit here:
> https://github.com/BIC-MNI/glim_image/tree/develop
>
>
> On 15-04-30 05:18 PM, Najmeh Khalili-Mahani wrote:
>
>> Hi,
>>
>> Is Glim_image totally extinct? I cannot find the documentation in Jon
>> Taylor's site and my memory of the command syntax has faded!
>>
>> I am looking for a way to do VBM, using surfstat-like design matrices.
>>
>> Thank you for sugegstions.
>>
>> Naj
>>
>> _______________________________________________
>> MINC-users at bic.mni.mcgill.ca
>> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
>>
>
>
> --
> Best regards,
>
>  Vladimir S. FONOV ~ vladimir.fonov <at> gmail.com
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
>

More information about the MINC-users mailing list