[Freesurfer] topological defect correction: Orig surface is right but WM surface is not right

Hello FS experts,   I have a MRI volume where there is regional hyperintensity at cortex and thus the segmentation takes that regional cortex as white matter. In other words, the segmentation process is too aggressive toward the pial surface.   I modifed the WM by manual erase across several slices and run -autorecon2-wm and -autorecon3. However, when I checked the results, I found the "original surface" is right (green color in the appended file; the tesselation is compatible with my manural editing) but the "white matter surface" is not updated (yellor color in the appended file; the surface is not adjusted, just like the old one before my editing).   Could anybody guide me how to solve this issue? Thank you very much, JoJo