[Freesurfer] Bad surfaces?

I reran our subjects from "final_surfaces" on through autorecon3 using 3.0.4 to correct for the pial surface crossover at the midline (as recommended by Bruce). Most things look fine, except that there are many portions of the surfaces that appear to "bunch" together. In one extreme example it appears that the white and pial surface cross. I've attached some jpgs to demonstrate what I mean. The close-up is r medial temporal area, while the other shows the "bunching" in the l superior portion, and at the midline. I understand that the crossing is not correct, but do the other surfaces look appropriate? Thanks.

-Derin