Create subject-specific volumetric ROI from FreeSurfer Desikan-Killiany-based annotations (aparc.annot)

Hi everyone

I ran fMRIPrep including the FreeSurfer pipeline.
For each subject, I want to convert annotations from the aparc parcellation (Desikan-Killiany atlas) to the T1w volume-space of the subject, to use as a ROI in MVPA (using cosmomvpa in matlab, with the GLM outputs from SPM12 based on the functional runs in T1w volume space).
In doing this, I want to combine the Fusiform, Parahippocampal and Infero-Temporal areas of each hemisphere and then combine across hemispheres to have one ROI per subject. When converting the ROI to the volume space, it would be useful I think if the ROI covers the full grey matter without crossing over into the white matter.

I’m unsure of all the correct steps I need to take to achieve this and hoped that somebody could help me out with that.
I want to make sure that everything is correctly registered for example. I’ve searched online, but could only find a couple of relevant tutorials that still leave me unsure of what to do exactly in my situation 1 2 3 .

The steps I think I need to take for each subject, but not sure if correct (order) and not sure about the specifics either:

  1. mri_annotation2label for each hemisphere
  2. do I need to perform some kind of registration on the labels?
  3. can I do mri_mergelabels or mri_binarize to combine all the labels?
  4. mri_label2vol for every label in every hemisphere, should I add —proj frac 0 1 0.01 so that the label will cover the full grey matter?
  5. do I need to perform some kind of registration now on the volumetric ROIs?
  6. how to combine all the ROIs into 1 ROI if that has not happened yet?

All help would greatly be appreciated!

1 Like