From OpenWetWare
Jump to: navigation, search
Brain picture
Beauchamp Lab

Previous step is Turning the raw data into AFNI BRIKs Next step is Making cross-subject averages with cortical surfaces


This page describes creating cortical surface models in FreeSurfer 6.0 (released in 2017) for use with AFNI and SUMA versions (current as of August 2017). These steps assume that you are in the correct directory and have already created an AFNI directory containing the anatomical images (see Turning the raw data into AFNI BRIKs). The Beauchamp Lab convention is to use a subject code for the experiment home directory and as a prefix for all files. For instance,

 set subj = XX
 cd /Volumes/data/BCM/{$subj}/afni

Step 1: Averaging Anatomical Scans Using AFNI

If you have only one anatomical data set, you can skip this step. FreeSurfer can handle multiple anatomies, but it is nice to register and average them in AFNI so we can have one average anatomy that we use for all purposes. This takes only a few minutes.

1. Register all the anatomicals to the space of the anatomical closest in time to the functional data (for instance, the last anatomy if the anatomies were collected before the functionals).

 3dAllineate -base series0003.nii -source series0002.nii -prefix ${subj}series2_RegTo3.nii -verb -warp shift_rotate -cost mi -automask -1Dfile ${subj}series2_RegTo3

Repeat this for as many anatomical datasets as you have.

2. Average all anatomicals into one dataset:

 3dmerge -gnzmean -nscale -prefix ${subj}anatavg.nii series0003.nii ${subj}series2_RegTo3.nii

Step 2: Run FreeSurfer

This step takes 12 or more hours. First, we must be in the home directory for the subject and set up the FreeSurfer environment variables to match.

 cd /Volumes/data/BCM/{$subj}
 setenv SUBJECTS_DIR `pwd`

Then, run FreeSurfer with the average anatomy created in the previous step. If you did not create an average anatomy, insert the name of the anatomical file to use.

 recon-all -all -parallel -subject fs -i afni/${subj}anatavg.nii 

Optional: If you are dissatified with the surface and you have a T2 dataset, this can be included in the recon as follows:

 recon-all -all -parallel -subject fs -i afni/${subj}anatavg.nii -T2 afni/T2anat.nii -T2pial

However, in our limited experience, this can push the pial envelope to approximately the midpoint between the WM boundary and the true pial surface.

Optional: For ECoG subjects, it is nice to have pial envelope (smooth ellipse shape of outer brain volume). This requires re-running FreeSurfer after the first run has completed. Also note that this step requires Matlab. Another option would be to use the AFNI program ConvexHull.

 recon-all -s fs -localGI

Step 3: Convert surfaces for use with AFNI/SUMA

This step takes a few minutes.

 cd /Volumes/data/BCM/{$subj}/fs
 @SUMA_Make_Spec_FS -NIFTI -sid fs

It is important to check the surface. To make this easier, the Beauchamp lab has a script to automatically load AFNI and SUMA and co-ordinate the viewers. Install the script like so:

 cd /Volumes/data/BCM/{$subj}  
 cp /Volumes/data/scripts/@ec .

Whenever data visualization is required, run the script like this


Page describing the @ec script

Other relevant pages:

  1. Finding Distances on the Surface
  2. Finding Closest node on the Surface
  3. How to make 3D print outs of cortical surface models

Older pages (FreeSurfer 5.3 + older versions of AFNI

  1. Preparation for Creating Cortical Surface Models
  2. Creating Cortical Surface Models
  3. Final touches and using Cortical Surface Models
  4. What If a Cortical Surface Model Exists Already
  5. What If Cortical Surface Model Looks Bad
  6. Creating Surface Averages of Functional Data
  7. SUMA
  8. Free Surfer

Very old pages that may not be useful

  1. Creating Standardized Surface Models
  2. FreeSurfer Standard Surface Models
  3. Caret