SUMA Tutorial #3: Analysis on the Surface

Overview

Surface-based analysis has many advantages, which we outlined earlier.

Once you have prepared the surfaces with @SUMA_Make_Spec_FS and @SUMA_AlignToExperiment, you can use the output with afni_proc.py. We will create a template in uber_subject.py, and then make the following changes:

  1. Insert a surf block after the volreg block;

  2. Remove the mask block;

  3. Remove any references to blur estimation in the regress block (i.e., regress_est_blur_epits and regress_est_blur_errts);

  4. Add the option -surf_anat which points to the surface anatomical image in the SUMA directory;

  5. Add the option -surf_spec which points to the spec files in the SUMA directory. If you want to analyze the individual subject’s surface, use the <subjID>_?h.spec files in the SUMA directories created by @SUMA_Make_Spec_FS. If you want to do a group analysis in standardized space, use the std.141.<subjID>_?h.spec files.

To create the preprocessing script, a typical afni_proc.py script would look like the following:

#!/usr/bin/env tcsh

# set subject and group identifiers
set subj  = sub-08
set gname = Flanker

# set data directories
set top_dir = /Users/ajahn/Desktop/${gname}/sub-08
set anat_dir  = $top_dir/anat
set epi_dir   = $top_dir/func
set stim_dir  = $top_dir/func

# 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 surf blur scale regress       \
-copy_anat $anat_dir/sub-08_T1w.nii.gz                         \
-dsets                                                         \
$epi_dir/sub-08_task-flanker_run-1_bold.nii.gz             \
$epi_dir/sub-08_task-flanker_run-2_bold.nii.gz             \
-tcat_remove_first_trs 0                                       \
-align_opts_aea -giant_move                                    \
-tlrc_base MNI_avg152T1+tlrc                                   \
-volreg_align_to MIN_OUTLIER                                   \
-volreg_align_e2a                                              \
-volreg_tlrc_warp                                              \
-surf_anat ~/Desktop/Flanker/FS/sub-08_T1w/surf/SUMA/sub-08_SurfVol+orig     \
-surf_spec ~/Desktop/Flanker/FS/sub-08_T1w/surf/SUMA/std.141.sub-08_?h.spec  \
-blur_size 4.0                                                 \
-regress_stim_times                                            \
$stim_dir/congruent.1D                                     \
$stim_dir/incongruent.1D                                   \
-regress_stim_labels congruent incongruent                              \
-regress_basis 'GAM'                                         \
-regress_censor_motion 0.3                                     \
-regress_motion_per_run                                        \
-regress_opts_3dD                                              \
-jobs 8                                                    \
-gltsym 'SYM: congruent -incongruent' -glt_label 1 Con-Inc \
-gltsym 'SYM: incongruent -congruent' -glt_label 2 Inc-Con \
-regress_reml_exec                                             \
-regress_make_ideal_sum sum_ideal.1D

This script would work for analyzing sub-08. To make it generalizable, we will make the same edits to the path as we did during the volumetric analysis.

We will also cut this code (which should be found around lines 218-219):

# set directory variables
set surface_dir = ${PWD}/FS/${subj}_T1w/surf/SUMA

And paste it around line 35, before we cd into the output directory.

To download the script that already has these edits, click here. You can then loop it over all of your subjects with the following code:

for i in `cat subjList.txt`; do
  tcsh SUMA_proc_MNI.sh $i;
  mv ${i}.results_SUMA $i;
done

Downloading the Surface Template

Just as with our volumetric analyses, we will need to render our results on a surface template. Navigate to the directory containing your subjects, and type the following:

afni_open -aw suma_MNI_N27.tgz
tar xvzf suma_MNI_N27.tgz

This will download and unpack the MNI_N27 surface template, which we can then view our results on.

Loading the First-Level Results

After the preprocessing script has finished, you will see a list of files output that have names similar to the ones you saw during the volumetric analysis. In this case, however, some of the files end in suffixes such as niml.dset. These are datasets that can be loaded into SUMA using the -input option.

To load the statistics for the left hemisphere for sub-01, for example, navigate to the directory sub-01/sub-01.results_SUMA and type the following:

suma -spec ../../suma_MNI_N27/std.141.MNI_N27_lh.spec -sv sub-01_SurfVol_Alnd_Exp+tlrc -input stats.sub-01.lh.niml.dset

You can then select the statistic sub-brik you would like to display, and change the p-value threshold accordingly.

Note

If the results have been normalized to MNI space, then use the code specified above. If you have processed the data in the subject’s native space and have not normalized the data, then you could use the subject’s spec file located in his corresponding surf directory, and change the surface volue to <subjName>_SurfVol_Alnd_Exp+orig.