[Freesurfer] trouble with bias correction

Some of the T1 scans I'm working with have pretty severe intensity inhomogeneity, with the signal dropping off significantly toward the sides of the head. This is not being adequately corrected and the segmentation is missing considerable portions of gray matter in those areas. I've played with some bias correction techniques but so far I have not been very successful.

In the attached screenshots, you can see the problem on the left side of the brain. The skull drops down significantly and in the ribbon file we lose all of the gray matter in that area. Is there a way to use control points to specify gray matter as well as white? Is there another obvious fix for this kind of problem?

Thanks! -Keith