Hello AFNI community.

I am suddenly having a problem with my 3dDeconvolve script seg faulting. I found one mention of this problem on the board where it was mentioned that taking out the -tout option fixed the problem, but that didn't work for me.

Some background.... All of our experiments are done in the same way so that we have a whole slew of fmri data files. These are all processed via a script that loops through the files (so all of them are processed exactly the same way). At this point I have 3 datasets per subject that consist of the three experimental conditions (active, passive, assisted). So next I run a deconvolve script that steps through the three datasets to do a regression. Here's the script....

#! /bin/bash

datasets="active assisted passive"

for i in $datasets
do
3dDeconvolve \
-input ./afni/$i-lp+orig \
-polort 1 \
-mask ./afni/full-mask+orig \
-concat ./afni/runs.1D \
-num_stimts 7 \
-stim_file 1 ./afni/full_stimulus.1D -stim_label 1 Motion \
-stim_file 2 ./movement/$i.dfile'[1]' -stim_base 2 -stim_label 2 Roll \
-stim_file 3 ./movement/$i.dfile'[2]' -stim_base 3 -stim_label 3 Pitch \
-stim_file 4 ./movement/$i.dfile'[3]' -stim_base 4 -stim_label 4 Yaw \
-stim_file 5 ./movement/$i.dfile'[4]' -stim_base 5 -stim_label 5 dS \
-stim_file 6 ./movement/$i.dfile'[5]' -stim_base 6 -stim_label 6 dL \
-stim_file 7 ./movement/$i.dfile'[6]' -stim_base 7 -stim_label 7 dP \
-fitts $i-fitts \
-full_first -fout -rout -tout -nobout \
-bucket $i-stats
done


This whole process works great on 9 of my twelve subjects. On the other two the 3dDeconvolve seg faults. Keep in mind all of this data is processed by the sames scripts in the exact same way.

One of the seg fault messages is shown here:
Program: 3dDeconvolve
Author: B. Douglas Ward, et al.
Initial Release: 02 September 1998
Latest Revision: 19 August 2004

../decontest: line 35: 5752 Segmentation fault 3dDeconvolve -input ./afni/$i-lp+orig -polort 1 -mask ./afni/full-mask+orig -concat ./afni/runs.1D -num_stimts 7 -stim_file 1 ./afni/full_stimulus.1D -stim_label 1 Motion -stim_file 2 ./movement/$i.dfile'[1]' -stim_base 2 -stim_label 2 Roll -stim_file 3 ./movement/$i.dfile'[2]' -stim_base 3 -stim_label 3 Pitch -stim_file 4 ./movement/$i.dfile'[3]' -stim_base 4 -stim_label 4 Yaw -stim_file 5 ./movement/$i.dfile'[4]' -stim_base 5 -stim_label 5 dS -stim_file 6 ./movement/$i.dfile'[5]' -stim_base 6 -stim_label 6 dL -stim_file 7 ./movement/$i.dfile'[6]' -stim_base 7 -stim_label 7 dP -fitts $i-fitts -full_first -fout -rout -tout -nobout -bucket $i-stats

I am terribly confused at this point. The exact script works on all 9 other subjects, just not on these two. I loaded up AFNI on another Linux machine (Slackware) and acheived the same results.

Does anybody have any ideas? I can only assume it's something about that particular data that is causing it, but I don't know what since, on inspection, the datasets look fine.

Any thoughts would be deeply appreciated.

iggy