startup & output

jip-glm (aka glm) is designed as a unix/linux program that requires that all necessary files are specified on the unix line at startup.  The general syntax is

glm  [GLM control file]  [options]

This program has few startup options.  The main option is "-p", which specifies that all signal-change files should use units of percent change by dividing the absolute signal changes, which are determined in GLM, by the DC component of the polynomial baseline description.

glm  glm.dat -p

This option is typical for BOLD or IRON fMRI, but it does not necessarily apply to ASL or to cases where the time series already has been converted to something other than signal (e.g., CBF, CBV).

Other input options

-h or -H        Get help, either the short version (-h) or the long version (-H).

-v or -V        Use verbose output.

-d [divisor]   Divide signal changes by divisor for output files.

-s                 Use very small memory allocation.  This generally isn't needed, as allocation already is

                    limited by default.  The program reads all input files in small bytes by reading a full time

                    series for a limited number of voxels each time.

-o [list]         Read an overlay list. Use this option in conjunction with -T or -S options to optimize one

                    or more parameters across a series of regions of interest. Up to 4 “-T” or “-S” options can

                    be used at the same time.

-T [ID]          Optimize the time constant for an event with the specified identification. Use this together

                    with the -o option. This adjust time constants for gamma or sigmoid functions (phMRI).

-S [ID]          Optimize the starting time for an event with the specified identification. Use this together

                    with the -o option.

-r                  Output the voxel-wise residue of the fit for each time point to a file named “residue”.

Output files

For each condition specified in the GLM control file, there will be 3 output files:

  1. P-value maps                      P-condition.nii               e.g., P-LR.nii
  2. T statistic maps                   T-condition.nii               e.g., T-LR.nii
  3. Signal change maps            S-condition.nii               e.g., S-LR.nii

Additionally, the file “average” contains the average value and chi-squared per degree of freedom for the fit.

The ratio will be the SNR.

Joseph B. Mandeville, Athinoula A. Martinos Center for Biomedical Imaging at MGH/MIT/Harvard