AFNI Class 05: 09/06/2019 [11/17 & 12/1/2025]

afni_proc.py for Resting and Nonlinear

using 11b modern [restingstate without freesurfer]

3dcalc -a /software/AFNI_25.2.06/TT_desai_dd_mpm+tlrc -expr 'amongst(a,152,170)' -prefix template_ventricle

We are replacing this with ... kmeans_ants.R

3dresample -dxyz 4 4 4 -inset MNI152_2009_Vent.nii.gz -prefix template_ventricle_4mm.nii.gz

afni_proc.py \

-subj_id ${subj} \

-blocks despike tshift align tlrc volreg blur mask scale regress \

-copy_anat ${subj}.nii.gz \

-dsets ${subj}rsfMRI_Run?.nii.gz \

-tcat_remove_first_trs 2 \

-align_unifize_epi local \

-copy_anat anatSS.${subj}.nii.gz \

-anat_has_skull no -align_opts_aea -cost nmi -giant_move -check_flip \

-volreg_align_to MIN_OUTLIER -volreg_align_e2a \

-volreg_tlrc_warp -tlrc_base MNI152_2009_template_SSW.nii.gz \

-tlrc_NL_warp -tlrc_NL_warped_dsets anatQQ.${subj}.nii.gz \

anatQQ.${subj}.aff12.1D anatQQ.${subj}_WARP.nii.gz \

-volreg_warp_dxyz 4 \

-blur_size 4 \

-mask_segment_anat yes \

-mask_segment_erode yes \

-mask_import Tvent template_ventricle_4mm.nii.gz \

-mask_intersect Svent CSFe Tvent \

-mask_epi_anat yes \

-regress_apply_mot_types demean deriv \

-regress_motion_per_run -regress_anaticor_fast \

-regress_ROI_PC Svent 3 \

-regress_ROI_PC_per_run Svent \

-regress_censor_motion 0.2 \

-regress_censor_outliers 0.05 \

-regress_make_corr_vols WMe Svent \

-regress_est_blur_epits -regress_est_blur_errts \

-regress_bandpass 0.01 0.1 \

-regress_run_clustsim yes

Which creates: proc.Rest.BP

Ways resting differs from task:

Censor level on frame displacement set to .2

1d_tool.py -infile dfile_rall.1D -set_nruns 2 \
-show_censor_count -censor_prev_TR \
-censor_motion 0.2 motion_${subj}

Removing Ventricular signal:

3dpc -mask mask_inter_Svent+tlrc -pcsave 3 \
-prefix rm.ROIPC.Svent.r${run} rm.det_pcin_r$run+tlrc
1d_tool.py -censor_fill_parent rm.censor.r$run.1D \
-infile rm.ROIPC.Svent.r${run}_vec.1D -write - \
| 1d_tool.py -set_run_lengths $tr_counts -pad_into_many_runs $run 2 \
-infile - -write ROIPC.Svent.r$run.1D

...

3dDeconvolve -ortvec ROIPC.Svent.r01.1D ROIPC.Svent.r01 \

-ortvec ROIPC.Svent.r02.1D ROIPC.Svent.r02 \

Bandpass filter set to .01 .1

1dBport -nodata $nt 2 -band 0.01 0.1 -invert -nozero >! rm.bpass.1D

1d_tool.py -infile rm.bpass.1D -pad_into_many_runs $run $#runs \

-set_run_lengths $tr_counts \

-write bpass.r$run.1D

...

1dcat bpass.r*1D > bandpass_rall.1D

...

3dDeconvolve -ortvec bandpass_rall.1D bandpass

Using 3dTproject to handle final analysis as data cleaning [note ran with and without WM resid]

3dTproject -polort 0 -input pb05.$subj.r*.scale+tlrc.HEAD \

-censor censor_${subj}_combined_2.1D -cenmode ZERO \

-dsort Local_WMe_rall+tlrc \

-ort X.nocensor.xmat.1D -prefix errts.$subj.fanaticor

Other Thoughts:

This path is primarily for graph theory work you can extract for working with graph metrics like igraph
- Can use 3dROIstats and 3dNetCorr
if you are doing seed based then replace the 3dTproject with the REML

3dDeconvolve -num_stimts 1 -stim_file 1 seed.1D -stim_label 1 seed

tcsh stats.REML_cmd

you can also consider doing ICA analysis for resting state consider using fsl melodic or Conn
you can also use GIMME for SEM style graph analysis

NL with task and @SSwarper:

first strip data:

basename `pwd` | xargs -I % @SSwarper -input %.nii.gz -base MNI152_2009_template_SSW.nii.gz -subid %

then you can add to your afni_proc.py the following:

-tlrc_NL_warp -tlrc_NL_warped_dsetsanatQQ.${subj}.niianatQQ.aff12.1DanatQQ.${subj}_WARP.nii