Problems with file regressor afni_proc.py 3dDeconvolve stage (EEG based regressors)

AFNI Message Board
1

Hi,

When I try running afni_proc.py using only the EEG regressor file. The function seems to replace most of the values in my EEG regressor.1D files with zeros and I don’t know why or how to fix it. I have attached some of the outputs I am getting and the warning messages I get while running the command. Can anyone tell me how to fix the issue? The output regressor used in the 3dDeconvolve stage seems to have only two TR points but my .1D EEG regressor files have the same number of puts as there are TR in the EPI fMRI dataset.

Here is the pipeline I am using:
foreach subj ()

echo $subj

set top_dir = /Volumes/Fortress

#3dWarp -deoblique -prefix $subj.EPI1_deob.nii $top_dir/SZ_Raw/$subj/impmo1.d+orig.HEAD
#3dWarp -deoblique -prefix $subj.EPI2_deob.nii $top_dir/SZ_Raw/$subj/impmo2.d+orig.HEAD
#3dWarp -deoblique -prefix $subj.MPRAGE_deob.nii $top_dir/SZ_Raw/$subj/mprage.d+orig.HEAD

set data directories

set anat_dir = $top_dir/EEG-FMRI/MRI/$subj/MPRAGE
set EPI_dir = $top_dir/EEG-FMRI/MRI/$subj
set stim_dir = /Volumes/Fortress/EEG-FMRI/MRI/$subj/timing_files
set stim_file_dir = /Volumes/Fortress/EEG-FMRI/MRI/$subj/EEG_regressors

run afni_proc.py to create a single subject processing script

afni_proc.py -subj_id $subj
-script proc.$subj -scr_overwrite
-blocks tshift align tlrc volreg blur mask scale regress
-copy_anat $anat_dir/T1w_$subj.nii
-anat_has_skull yes
-mask_epi_anat yes
-tlrc_base MNI152_T1_2009c+tlrc
-dsets
$EPI_dir/attVis1/attVis1_$subj.nii
$EPI_dir/attVis2/attVis2_$subj.nii
$EPI_dir/attAud1/attAud1_$subj.nii
$EPI_dir/attAud2/attAud2_$subj.nii
-tcat_remove_first_trs 0
-volreg_align_to MIN_OUTLIER
-volreg_align_e2a
-volreg_tlrc_warp
-blur_size 6.0
-test_stim_files no
-regress_no_stim_times
-regress_use_stim_files
-regress_extra_stim_files
$stim_file_dir/Alpha_VisVis.1D
$stim_file_dir/Alpha_VisAud.1D
$stim_file_dir/Alpha_AudAud.1D
$stim_file_dir/Alpha_AudVis.1D
$stim_file_dir/Alpha_response_VV.1D
$stim_file_dir/Alpha_response_AA.1D
-regress_extra_stim_labels Alpha_VV Alpha_VA Alpha_AA Alpha_AV resAlp_VV resAlp_AA \

-regress_stim_times \

$stim_dir/VisVis_all.1D \

$stim_dir/VisAud_all.1D \

$stim_dir/AudAud_all.1D \

$stim_dir/AudVis_all.1D \

-regress_stim_types AM1 AM1 AM1 AM1 \

-regress_stim_labels \

VisVis VisAud AudAud AudVis \

    -regress_basis 'dmBLOCK(1)'                                  \

-regress_est_blur_errts \

-regress_opts_3dD					\
	-GOFORIT 2				\
    -regress_apply_mot_types demean deriv                             \
	-regress_motion_per_run						  \
	-regress_censor_motion 0.5 				  \
	-regress_censor_outliers 0.1				  \
-regress_apply_mask  						\
-regress_run_clustsim no					  \
-regress_no_fitts 						\
-out_dir $EPI_dir/$subj.results.scploc_EEGinformed_2							                           
 

end

Thank you,
Dalton

Screen Shot 2021-12-02 at 9.27.44 AM.jpg
Screen Shot 2021-12-02 at 9.27.44 AM.jpg2264×700 301 KB

2

Hi Dalton,

It might be helpful to see the warning messages. But either way, it looks like the stim files are being interpreted as timing, rather than as pure regressors.

Since the -regress* options for that look okay to me (note that you can alternatively use -regress_stim_times … -regress_stim_types file file …), it seems like there is an alternate error in the script.

Can you follow up with 3 items?

  1. The warnings that you are concerned about, hopefully along with the command that produces them.
  2. The 3dDeconvolve command, at least with the -stim* options.
  3. The afni_proc.py command that is commented at the very end of the proc script, to be sure of what afni_proc.py is seeing.
  • rick