Hi AFNI Experts,

I would like to run 3dttest++ to look at the difference between two groups on a statistical contrast map ConditionA>ConditionB. I would like to restrict the search for significant voxels using an ROI for the amygdala, and also to correct for multiple comparisons using 3dClustSim.

I am using Peter's blog as a guide: blog.cogneurostats.com/?p=196 but I am very stuck at 3dFWHMx. I have not used afni_proc.py, and am having great difficulty understanding it to try to use that to guide me.

I have individual subject ROI masks manually created using Draw Dataset from the extracted contrast maps from 3dDeconvolve, which have been spatially normalized and smoothed already [ROImask_Lamy+tlrc, ROImask_Ramy+tlrc], and the residual error file from 3dDeconvolve -errts [residual_error.fit+orig].

To get the average group mask, I first combined the left and right amygdala using: 3dcalc - a ROImask_Lamy+tlrc -b ROImask_Ramy+tlrc -prefix ROImask_BilatAmy -expr '1*a + 2*b'

Next, since the masks were already in standardized space, I used 3dMean to combine them to create Avg_ROImask_BilatAmy+tlrc.

To apply the mask to 3dttest++ I believe I just use the flag -options -mask Avg_ROImask_BilatAmy+tlrc.

However, I am not sure how to move onto multiple comparison correction in terms of estimating the blur with 3dFWHMx for each of the individual subject ROImask_BilatAmy+tlrc files.

They are based on spatially normalized and smoothed data (i.e., are in tlrc space and smoothed with -1blur_fwhm 6), so do I not need to run 3dFWHMx and for 3dClustSim use -fwhmx 6 6 6, or do I need to transform the masks somehow, for example with 3dfractionize, and in that case, how?

I would assume I want the mask in the same dimensions as the error file (residual_error.fit+orig), but when using:
3dfractionize -template residual_error.fit+orig -input ROImask_BilatAmy+tlrc -clip 0.2 -preserve -prefix Frac_ROImask_BilatAmy

I get an error:
++ 3dfractionize: AFNI version=AFNI_2011_12_21_1014 (Jun 11 2013) [32-bit]
++ Authored by: RW Cox
** -preserve option requires short- or byte-valued input dataset!

If I exclude -preserve the dataset is written, but when I look at the ROI mask on the original anatomical (run1-T1+orig), it is not in the correct location of the amygdala.

I am not sure how to proceed and am hoping someone may be able to offer some guidance, and relatively quickly (I am under an extremely tight and crutial deadline).

Thanks so much in advance!