Hi Roser,
it's hard to judge accuracy from a single slice. The cc segmentation adjusts for the angle of the head, which can make any single sagittal slice look wrong as the cc is cutting through it at an angle. Try looking at the segmentation in other views. Horizontal can be particularly helfpul. If you are still convinced it is wrong you can either edit it manually in tkmedit, or send the subject to us and we'll take a look.
cheers, Bruce
On Mon, 22 Mar 2010, Roser Sala wrote:
Hello,
We are running FreeSurfer in order to obtain Callosal volumes from pathologic brain.
Segmentation volumes (aseg.mgz) seem 'not bad', but specially in the Callossum there are some misclassifications (attached image with an example).
Which would be the best way to correct this? Do you recommend us to add some control points? we are not sure if adding control points will affect only the surface reconstruction, or will help aslo to correct segmentation, which is our purpose.
Many thanks in advance
- Roser -