[Freesurfer] control points and white matter edits

Hi all,

I'm having trouble getting accurate white and pial surfaces in inferior temporal regions on a few lower quality scans with significant intensity variation. I've added hundreds of control points and drawn in white matter in these regions when control points failed to recover unlabeled cortex, but when re-running the scripts with either -autorecon2-wm, or -autorecon2-cp, followed by -autorecon3 the surfaces are still not accurate. Are there other interventions I can to to get more accurate surfaces? If you manually add white matter why will the regenerated white matter surfaces not always extend around the hand drawn parts?

Thanks, Paul