On concatenating resting runs

Hello,

I have two resting state runs for the same subject, but when I pass them to afni_proc using -dsets rest* neither of them have enough TRs after censoring to run 3dDeconvolve. There’s a chance I’d have enough TRs if 3dDeconvolve were run on a concatenated time series. (this is a generic description - we have several subjects in such scenario and the number of TRs left after censoring varies by subject).

How would you go about concatenating these files before running afni_proc (e.g. scaling, time shifting, polort regression)? Or are there any flags in afni_proc I’m missing that would take care of that (below, but mostly example 11)? In general, is it even a good idea to do this?

Thanks,

Gustavo

===

afni_proc.py -subj_id $subj
-blocks despike tshift align volreg blur mask regress
-copy_anat ${SUMA_dir}/${subj}_SurfVol.nii
-out_dir $subj_dir/$subj.rest.subjectSpace.results
-script $subj_dir/rest.proc.subjectSpace.$subj
-anat_follower_ROI aaseg anat ${SUMA_dir}/aparc.a2009s+aseg.nii
-anat_follower_ROI aeseg epi ${SUMA_dir}/aparc.a2009s+aseg.nii
-anat_follower_ROI FSvent epi ${subj_dir}/FT_vent.nii
-anat_follower_ROI FSWe epi ${subj_dir}/FT_WM.nii
-anat_follower_erode FSvent FSWe
-dsets ${subj_dir}/rest*+orig.HEAD
-tcat_remove_first_trs 3
-volreg_align_to MIN_OUTLIER
-volreg_align_e2a
-regress_ROI_PC FSvent 3
-regress_make_corr_vols aeseg FSvent
-regress_anaticor_fast
-regress_anaticor_label FSWe
-regress_censor_motion 0.2
-regress_censor_outliers 0.1
-regress_apply_mot_types demean deriv
-regress_est_blur_epits
-regress_est_blur_errts
-regress_run_clustsim no

Hi, Gustavo–

Well, if you are running out of degrees of freedom/time points in each time series without even bandpassing, then it seems like these subjects should probably not be used-- there is just too much motion, presumably.

–pt

Paul,

Thanks for the feedback. If I were to use those scans just to analyze the effects on the overall group results, is there a way to specify the run concatenation in afni_proc, or through some other combination of AFNI tools to perform the appropriate pre-processing?

G

Against best advice… You could use 3dTcat[/url] in combination with (likely) [url=https://afni.nimh.nih.gov/pub/dist/doc/program_help/3dDetrend.html]3dDetrend[/url] or some creative use of [url=https://afni.nimh.nih.gov/pub/dist/doc/program_help/3dSynthesize.html]3dSynthesize.

I might suggest that a (perhaps) slightly less evil approach would be to make your motion threshold more liberal (-regress_censor_motion).