The GLM implementation for SRTM is similar to that for fMRI. If you know jip-glm, this will be familiar.
- Create a separate directory to hold the contents of the GLM analysis. This isn't a requirement, but you will find that GLM analyses can create a lot of files. Also, you may want to have several separate GLM analyses, with each grouped in its own directory.
- This GLM analysis requires one top level "control file" that specifies various options and also lists the runs (data files) to be analyzed by GLM. The control file format is here. At the top of the file, there are keyword-delimited parameters followed by arguments (e.g., "reference-region file.ovl"). The keywork "runs" or "runs:" indicates the end of this section and the beginning of a list of runs. Subsequent lines are:
[PET data file name] [GLM description file name] [table file name]
- Each run (or "scan" or "data file") requires a "GLM description file" (typical extension is .glm) that determines the specific analysis within SRTM. Additionally, a mandatory "table file" is associated with each run. The table file contains a series of columns with a header for each column, plus one entry for each column for each frame. The total number of lines in the file should equal the number of frames plus 1 (for the header). The first column must contain the length of each frame in units of seconds. Additional columns, if they exist, can contain anything and can be used, in principle, as nonparametric regressors that specify some specific form for a change in binding potential, for instance.
- Run the GLM, generally using a single argument to convert absolute signal changes to % changes:
where "glm.dat" is the name of the control file in this case. Alternatively, one might want to first adjust
GLM parameters using the -t or -T options.
- Display the time series and all maps using jip-display (a.k.a, xd):
"xd -t srtm.dat &" or "xd &" as shown here
- When "glm" was executed in step 4, it created a hidden file named ".display" with default contents "xd -t glm.dat". If no arguments are given to "xd", and if this file exists on the current directory, then the arguments within the file will be used. This is why one can type simply "xd" in step 5. Consider modifying file ".display" using an editor to include whatever arguments are useful. Then, when you revisit this directory many months later, you can simply type "xd" to bring up the analysis. Example:
xd -t glm.dat -e /space/1/me/template/anatomy.nii -o /space/1/me/template/overlay-list.dat