Question on TENT function in 3dDeconvolve

Hello,

I’ve used TENT function to make a visual-stimulation related regressor in 3dDeconvolve.

The script is as follows:

set basis = ‘TENT(0,12,7)’

3dDeconvolve -input $work_dir/pb04.$subj.r0*.scale+tlrc.HEAD
-censor $work_dir/motion_{$subj}_censor.1D
-mask $work_dir/full_mask.$subj+tlrc
-polort A -float
-allzero_OK
-num_stimts 7
-stim_times 1 $root_dir/file_regressors/{$subj}_onStim.txt $basis
-stim_label 1 onStim
-stim_file 2 motion_demean.$subj.1D’[0]’ -stim_base 2 -stim_label 2 roll
-stim_file 3 motion_demean.$subj.1D’[1]’ -stim_base 3 -stim_label 3 pitch
-stim_file 4 motion_demean.$subj.1D’[2]’ -stim_base 4 -stim_label 4 yaw
-stim_file 5 motion_demean.$subj.1D’[3]’ -stim_base 5 -stim_label 5 dS
-stim_file 6 motion_demean.$subj.1D’[4]’ -stim_base 6 -stim_label 6 dL
-stim_file 7 motion_demean.$subj.1D’[5]’ -stim_base 7 -stim_label 7 dP
-gltsym “SYM: onStim”
-glt_label 1 onStim
-iresp 1 iresp.$subj.stimOn
-fout -tout -x1D X.xmat.onStim.1D -xjpeg X.onStim.jpg
-x1D_uncensored X.nocensor.xmat.onStim.1D
-bucket $work_dir/statsVis.$subj

{$subj}_onStim.txt file represents the stimulation onset timing in seconds. The final X matrix, X.nocensor.xmat.onStim.1D is 1746x48 matrix, which 36 to 42 columns represent the regressors for TENT function. We figured out the regressors for TENT function in MATLAB with each column corresponding to each regressors in TENT function (7 columns in total).
The resulting figures are attached. It seems weird since the regressors of TENT function should be the finite set of peaks with same amplitude in my knowledge. Stimulation onset timing file ({$subj}_onStim.txt) is not exactly integer number. For example, it is 4.02 8.03 12.03 20.03 … . Is the weird looking regressors due to its not-integer time? I’d be very appreciated to the answer. Thank you so much.

Are you sure the TR in the dataset is not 20 ms, for some reason,
and not 2 s? What is the output of:

3dinfo pb04.$subj.r01.scale+tlrc |& head -n 22

Also, what is the complete text output to the screen when
that command is run?

In any case, it is problematic for the stim times to be all
slightly off a TR, as that can end up scaling the initial
beta weight. Consider Example 7a or 7b from the output
of “timing_tool.py -help”. But this is a separate concern.

  • rick