(continued)
So, to convert the F-stats to signed z-scores: Let's suppose that the output
bucket dataset, stim01.bucket+orig, contains the following 2 sub-bricks (these
two sub-bricks were created using the -glt option):
3dinfo stim01.bucket+orig
etc. etc.
-- At sub-brick #14 'Area LC[0] coef'
-- At sub-brick #15 'Area F-stat'
statcode = fift; statpar = 1 247
etc. etc.
The following 3dcalc command converts the F-stats in sub-brick #15 to z-scores,
and then multiplies by the signs contained in sub-brick #14:
3dcalc -a stim01.bucket+orig'[14]' -b stim01.bucket+orig'[15]' \
-expr "(step(a)-step(-a))*fift_t2z(b,1,247)" -fscale -prefix temp
Note that the F-stat num and den dof (in this case, 1 and 247) must be the same
as those from the 3dinfo listing.
Now, to label the sub-brick, and attach it to the bucket dataset (optional),
use the following commands:
3drefit -substatpar 0 fizt -sublabel 0 z-score temp+orig
3dbucket -glueto stim01.bucket+orig temp+orig
Doug Ward