GLM control file (fMRI)

A GLM control file is the "top level" that defines the GLM to be used for analysis.


The first several lines describe options to used in the analysis, such as the number of terms describing baseline signal drift, the conditions to be tested, the hemodynamic impulse response model, as so on.  Required options include a list of "conditions", which specify statistical tests and also parameter estimation for output.


After the key-word delimited top section of the file, use the keyword "runs" or "runs:" to denote that subsequent lines in the file should be interpreted as runs.


Following the "runs:" line, all subsequent lines are interpreted as containing either 2 or 3 entries. The first 2 entries are required and are the name of the 4-dimensional data file and a file that specifies the stimuli for that run. Optionally, one may also include a table file that has entries for each time point and contains things (eye position, blood pressure, etc) that can only be entered explicitly in this way.


   An example GLM control file follows:


              baseline-terms 3           # include 3 terms in baseline polynomial
              IRF BOLD-rat               # use an IRF appropriate for BOLD in rats

              normalize-BOLD           # global scale factor across BOLD runs
              convolve-table 3 4        # convolve table events “3” and “4” with the IRF

              conditions 1 2 12 1-2    # see conditions page in documentation
              runs:                             # this indicates that following lines will list runs
              ../010/data.nii ../010/stim.glm  ../010/table.dat
              ../011/data.nii ../011/stim.glm  ../011/table.dat
              ../012/data.nii ../012/stim.glm  ../012/table.dat


The following keywords are defined for use in the GLM control file above the “runs:” line.


    Keyword                     argument        default value           purpose


    baseline-terms            [integer] (0-4)        3                       baseline polynomial
    notify                            none                                              verbose output
    multiple-comparisons  [integer # of resels]                       correct p-values
    normalize-BOLD        none (default = no normalization)                                             

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