Hi Hema,
To generate a set of voxels that should show
a reasonable correlation with the covariate,
use the t-stat of the covariate from 3dttest++
for thresholding, not the t-stat of the mean.
The 3dttest++ results look the same as the
3dTcorr1D results showing the voxelwise
correlation of the covariate to the contrasts
(just for comparison).
Does this seem reasonable?
- rick