Hi Jon
the parcellation labels are usually pretty accurate, particularly since gyral borders are pretty arbitrary. I wouldn't touch them unless you see a dramatic and obvious failure (e.g. precentral gyrus labeled as post-central). This happens pretty rarely though Bruce
On Fri, 31 May 2013, Jonathan Holt wrote:
Hi all,
We are using the parcellation of Desikan-Killiany
https://surfer.nmr.mgh.harvard.edu/fswiki/CorticalParcellation
and creating labels for each structure in each hemisphere with:
mri_annotation2label --subject ${SUBJ} --hemi ${HEMI} --outdir ${SUBJ}/${SUBJ}_annotate
and making matlab text readable (via matlab load command) thickness files with:
mris_convert -c ${HEMI}.thickkness ${HEMI}.white ${HEMI}.thickness.asc
Then we are reading in the ascii thickness files and the labels and creating histograms (unit area) for each structure for each participant, essentially following:
https://surfer.nmr.mgh.harvard.edu/fswiki/UserContributions/FAQ#Q.HowcanIma
keahistogramofcorticalthickness.3F
We've dropped the zeros but also dropping values of thickness identically = 5mm.
The question is: How important are parcellation edits, that is how accurate are the generated segmentation labels, and should we be dedicating effort to ensuring they're accurate and possibly editing them if not (if so, what should be used as a guide/template?) jon