I’m working with MRI data from individuals with multiple sclerosis (MS). I preprocess their scans using lesion filling in FSL, based on lesion masks prepared from FLAIR images.

However, for some subjects with extensive lesions—particularly large black holes and/or juxtacortical lesions—the surface reconstruction remains inaccurate, even after applying lesion filling (and of course without it as well).

I would like to manually correct the reconstruction in these problematic areas. I’ve tried adding control points directly inside the black holes (!) and also editing the wm.mgz by manually adding voxels, but it seems that directly editing the surface mesh would probably be more effective and less cumbersome. Is this feasible? Are there established tools or workflows that allow manual surface mesh correction in this context?

Please note that in the example provided, there is an area where, despite the correct wm.mgz mask, the surface does not follow the mask boundary.

Alternatively, I would appreciate any suggestions on other methods or pipelines that might be worth trying to improve surface reconstruction in such challenging cases.

Thank you!