Autorecon3

The third set of recon-all steps does an automated parcellation of the gyri and sulci of the cortical mesh (as described in this paper. It is invoked by calling recon-all -autorecon3. A shell script, autorecon3.sh has been created to simplify this step.

Run autorecon3.sh on a participant

autorecon3.sh 501

Some while later (about 90 minutes locally), the script will terminate and you can check out how well the parcellation turned out using tksurfer.

tksurfer 505_001 lh pial

Then File > Label > Import Annotation > Browse… > select corresponding ?h.aparc.annot file in that subject’s /label/ folder.

Here, the ‘?’ is a placeholder character, standing in for either ‘l’ or ‘r’, meaning you can select ‘rh.aparc.annot’ or ‘lh.aparc.annot’ (i.e., right or left hemisphere)

If you loaded the lh pial surface, you need to load the lh.aparc.annot. Similarly, if you loaded the rh pial surface, you need to load the rh.aparc.annot file.

Alternatively, you can load the cortical mesh and the labels simultaneously.

tksurfer -annot aparc 505_001 lh pial

If you select the wrong aparc.annot (e.g., selecting lh.aparc.annot when you have loaded the rh pial surface mesh, or if you load up the annotation table for a different participant), you will find your mesh gets a very interesting-looking, but utterly useless psychedelic color scheme.

Though they seem trivial, the colors assigned to each of the triangle mesh faces are used by Freesurfer to organize the anatomical labels assigned to every single point in the brain (i.e., every point assigned the same color is grouped in the same brain region). Freesurfer has tools (tksurfer) that allow you to go in and manually edit the parcellation if there is something ‘off’ about the label. For example, one of the sulci (the valleys in between the gyri) might extend too far up and over one of the adjacent gyri for your liking. Remember that Freesurfer uses computer algorithms that automate this parcellation. Though they do an extremely good job considering, there may be cases where your eye can improve upon the job that these algorithms have done. A tutorial on how to manually edit these parcellations using tksurfer can be found here. Aside from making minor fixes to the parcellations, you can use the procedure described in this tutorial to define anatomical regions of special interest to you, such as making a custom template that has definitions of the visual word form area or primary auditory cortex, based on some functional analyses you might have performed on some dataset.