Anterior/Posterior Commissure Segmentation

[sasaigoli05]

Hello - I just had a quick question regarding whether FastSurfer segments the Anterior Commissure and Posterior Commissure. I know that as of 2020, Freesurfer does not, so I just wanted to check.

[m-reuter]

We are currently working on including a corpus callosum module which also segments AC and PC. But this will probably not be in the upcoming release, but the one after that.

[dkuegler]

FastSurfer currently does not segment the anterior or posterior commisure. BUT

FastSurfer-HypVINN, which is already part of the development version of FastSurfer (see #459), does segment the anterior commisure as a dedicated structure. The posterior commisure is too small for reliable results (DICE etc.).

Note, that the development version of FastSurfer is targeted to developers and may have bugs. The next release of FastSurfer will have a user-safe version validated by us.

[sasaigoli05]

Thank you- I was hoping to identify slices parallel to the AC/PC plane. To confirm- that would not necessarily be possible in this case?

[dkuegler]

You could just use the talairach transforamation for this purpose. FastSurfer also calculates the talairach transformation similar to to FreeSurfer (SUBJECTS_DIR/subject_name/mri/transforms/talairach.lta). Or similarly, the upright transform contains similar infos /transforms/cc_up.lta.

[sasaigoli05]

That makes sense! So to use the registered image, the transformation matrix from the talairach.lta needs to be applied to the original MRI data? Or does the FastSurfer pipeline naturally complete this step?

[dkuegler]

FastSurfer does not apply the talairach transform in its pipeline, you need to do that manually.