Hi,

I'm using Freesurfer to segment MRI from patients with cortical lesions. The automatic segmentation does not work correctly, so I edited the wm.mgz manually and filled the missing white matter according to the tutorial. After this I ran recon-all with the -autorecon2-wm parameter. In some cases this works fine, but in a few cases the segmentation is still not correct. The wm surface forms a hole at the lesion despite the fact that I have manually filled the wm.mgz at that region. The pial surface forms an even deeper hole at the same spot, so that the surfaces cross.

I have filled the gap from end to end but still the segmentation does not work correctly. What would be the correct approach to this problem?

Thanks in advance, Martti Ilvesmäki