intergroupavg-sess - computes random effects analysis between 2 groups or subject populations (i.e. controls vs. patients)


intergroupavg-sess -analysis <analysisname> -contrast <contrastname> -group1 <group1name> -group2 <group2name> -intergroup <intergroupname>-space <spacename>


Positional Arguments


Required Flagged Arguments

-analysis analysisname

session-level functional analysis name

analysis created by and used with selxavg-sess on each individual session

-contrast contrastname

contrast used in the random effects avg

contrast created by and used when running isxavg-re-sess If the contrast is multivariate, then sig and minsig maps will be produced

-group1 groupname

name of group1 (positive)

group name given when calling isxavg-re-sess. This refers to the group that will be "positive

-group2 groupname

name of group2 (negative)

group name given when calling isxavg-re-sess. This refers to the group that will be "negative

-intergroup intergroupname

name of intergroup average

name given to the intergroup average. This name is used in subsequent commands just as one would use the session id or the group name

-space spacename

space in which to average (tal, sph)

Space in which {\em isxavg-re-sess} was run. This is the space that each individual was transformed to so that all individuals across all groups are registered. Options are: tal (for talairach), sph (for spherical surface-based), or the name of a region-of-interest (created with func2roi-sess

Optional Flagged Arguments

-polarity +1 or -1

only consider voxels with avg > 0 (or < 0)

-spacedir dir

default is spacename

-hemi hemisphere

with sph space <lh rh>

-umask umask

set unix file permission mask


don't run, just generate a script


print version and exit



intergroupavg-sess is a program for intergroup averaging. The individual groups are assumed to have been averaged using a random effects model (see isxavg-re-sess). Inter-group statistics are computed based on a t-statistic. The average and variance for each group are computed according to a random effects model. To run the random effects averaging, the user chooses a contrast which is used to collapse the hemodynamic responses across time and condition to yield a single number for each subject. The group average will be the average of this number over all the subjects. The group variance will be the variance of this number across all the subjects in the group.


Example 1: Whole Processing Stream

Say you have 10 subjects in 2 groups (A and B) with sessids: subject1a, ..., subject5a, subject1b, ..., subject5b.

Create sessid files called a.sid and b.sid and sessdir file sess.dir. Let the analysis name be "main. The contrast you are interested in is "up-vs-down.

Run selxavg-sess on all subjects:

Resample them into talairach space:

Run random effects group averaging for each group:

Run intergroup averaging

View the results on the talairach volume:

Example 2: Spherical

intergroupavg-sess -analysis <analysisname> -contrast <contrastname> -group1 controls -group2 patients -intergroup controlsVpatients-space sph -spacedir <outdir from sphsmooth-sess

Example 3: Volume/Talairach

intergroupavg-sess -analysis <analysisname> -contrast <contrastname> -group1 controls -group2 patients -intergroup controlsVpatients-space tal -polarity 1

performs a between-group average but only considers voxels that are <1 for the given contrast for each of the two groups (i.e. only considers voxels that were "red" in the given contrast for controls and for patients, and only uses these voxels in the between-group contrast



See Also



FreeSurfer, FsFast

Methods Description



Reporting Bugs

Report bugs to <>


DougGreve PhD

intergroupavg-sess (last edited 2008-04-29 11:46:05 by localhost)