We are analyzing data from our collaborators and are running into some difficulties with the accuracy of the pial and wm boundaries. The wm.mgz appears to follow the actual WM, independent of the wm surface boundary. Attached is an example showing two types of problems.

1. the WM surface is inaccurate regardless of the number of control points used.

2. The pial surface is inaccurate, missing a bunch of the sulcal and gyral patterns of the left parietal lobe (see area around cursor).

For completeness, I have also included the wm.mgz at the same slice. 

I should mention that these problems has been occurring in multiple subjects.

Any suggestions??

Thanks

Corinna