External Email - Use Caution
Hello All,
I used the default freesurfer command (recon-all) to generate the annotated surface file for subjects. However, the results in some subjects seem to have an obvious parcellation error. As you can see in the below graph, the area pointed by the red arrow is labelled as S_precentral-inf-part; however, we think it should be S_central. (In this particular subject, the parcellation is totally fine on the left hemisphere. The problem is only on the right hemisphere as you can see below)
We adopted a solution from this answer: https://secure-web.cisco.com/1qxgC1JMOwKWxmSccucR1CmKcgaAUMcy4N4o0VjYQs9ezRt...
We specified three manual labels: S_central, S_temperal_sup, and S_pericallosal, to align with atlas labels and generate a new registered surface file. Adding -nonorm option made the result even better.
S_central.label
S_temperal_sup.label
S_pericallosal.label
Command #1:
mris_register -L S_central.label <atlas_gcsa_file> S_central
-L S_temperal_sup.label <atlas_gcsa_file> S_temperal_sup
-L S_pericallosal.label <atlas_gcsa_file> S_pericallosal
-curv -nonorm sub-xxxxxx/surf/rh.sphere /freesurfer/7.3.2/average/rh.average.curvature.filled.buckner40.tif rh.sphere.reg.new
<atlas_gcsa_file> here is rh.destrieux.simple.2009-07-29.gcs
Command #2:
mris_ca_label sub-xxxxxx rh rh.sphere.reg.new <atlas_gcsa_file> rh.new.aparc.a2009s.annot
The parcellation looks a lot better now but it is still not perfect (for example, look at the area pointed by the red arrow)
Now, my questions are:
1.
Does our solution make sense? 2.
Are there any other better methods to correct the parcellation error?
Thank you!
Best wishes,
Junxia