FreeSurfer editing

Massive pial surface problem caused by sinus being bright in the T1. Very typical FreeSurfer problem. The new T2pial flag generally fixes things, but sometimes errors still persist.

Another sinus-induced problem. Very typical. Be on the lookout!

Common problem in more frontal coronal slices where the dura confuses the pial surface placement

Left is visualization of pial surface reconstructed with no -T2pial option. Right is with -T2pial. Notice the bumpiness/jaggedness of the left surface. (Not sure how much of a problem this is.)

Black is pial surface from initial FS reconstruction; yellow and cyan is pial surface after manual edits. Notice the pial surface juts too far into the sinus.

Same as previous image but with T2 data as the underlay

Cyan and red are cases of T2 being super bright (i.e. T2/T1 is very large) and T2 being super dark (i.e. T1/T2 is very large), respectively. Super bright T2 is CSF; super dark T2 is sinus. This is why the T2 is so useful for checking the accuracy of your cortical surface reconstructions.

Left is FS 7.4.1; right is FS 8.1. Notice catastrophic failure on the red contour in FS 8.1. It is not clear why this poor outcome occurs. In any case, it is a common problem (not just for FS 8.1), so be on the lookout for this in general!

The T1 anatomical data (using our acquisition) commonly has dropout in difficult places. Shown crosshair is near orbitofrontal cortex. Another location is near ear canal. This dropout leads to poor FreeSurfer surfaces. One approach is to resign oneself to not getting good surfaces in these regions (instead of trying to painstakingly correct these regions). Notice the "mean" image (which is fMRI signal intensity) is typically really dark (signal dropout) in these regions, so you aren't going to get any fMRI signals in these regions anyway...