3dDeconvolve 3dTproject


Sorry for my disturbance which is keep rising for a while and thanks for your help. I had a question about error time series, fitts and stats. I perform RS-FMRI analysis which is supported by uber scripts. I wanted to ask some part of my afni_proc.py generated script. Which is at following.

run the regression analysis

3dDeconvolve -input pb04.$subj.r*.blur+tlrc.HEAD
-mask full_mask.$subj+tlrc
-ortvec ROI.WMe_rall.1D ROI.WMe
-ortvec mot_demean.r01.1D mot_demean_r01
-ortvec mot_deriv.r01.1D mot_deriv_r01
-polort 2
-num_stimts 0
-fout -tout -x1D X.xmat.1D -xjpeg X.jpg
-fitts fitts.$subj
-errts errts.${subj}
-bucket stats.$subj

– use 3dTproject to project out regression matrix –

3dTproject -polort 0 -input pb04.$subj.r*.blur+tlrc.HEAD
-mask full_mask.$subj+tlrc
-ort X.xmat.1D -prefix errts.${subj}.tproject

This is the part that I want to ask. There is no output named fitts and stats. Only from tproject errts. I read at forum it is happening because of the x1D_stop So when I perform further analysis I think something is not correct because I use errts for correlation. My afni_proc.py is at the following;


script generated by the command:

afni_proc.py -subj_id subjLast -script proc.sub1 -scr_overwrite -blocks \

despike tshift align tlrc volreg blur mask regress -copy_anat \

/mnt/f/TEZ/Bipolar/female/${subj}/anat/${subj}_T1w.nii -dsets \

/mnt/f/TEZ/Bipolar/female/${subj}/func/${subj}_task-rest_bold.nii \

-tcat_remove_first_trs 4 -align_opts_aea -cost lpc+ -giant_move \

-volreg_align_to first -volreg_align_e2a -volreg_tlrc_warp -blur_size \

6.0 -mask_apply epi -mask_segment_anat yes -mask_segment_erode yes \

-regress_ROI WMe -regress_apply_mot_types demean deriv \

-regress_motion_per_run -regress_est_blur_errts -regress_RSFC \

-regress_bandpass 0.01 0.1


Hi Abdullah,

You are correct that there should be no fitts and bucket output, only the tproject errts, due to -x1D_stop. The errts.tproject is the correct file to proceed with, as it has had the given noise components projected out.

It is still called errts to be consistent, and because it is the residual time series of a linear regression. But it is indeed what might be used as input to a correlation analysis.

  • rick

Thanks for this fast response.

Can I ask one more which is not about this topic ? You can see my afni_proc.py it uses regress_bandpass and regress_RSFC both. Will bandpassing cause any harm on fALFF ? I asked this before and I understood that way. Is that the correct way to use ?


Hi Abdullah,

I believe that is correct to do. In that case, the -regress_bandpass frequency ranges should be passed to the RSFC step.

Note that the bandpassing would NOT be done in the typical regression step in this case, since RSFC will need the bands present.

  • rick