Missing sulci in the Mindboggle-101 brains

[binarybottle]

I used mindboggle to extract folds from the (manually labeled) Mindboggle-101 brains. A label boundary defined as a sulcus in the DKT labeling protocol must be in a mindboggle-extracted fold to be detected as a sulcus. When Elias Chaibub Neto analyzed shape measures in these brains, he found that all callosal sulci were missing from all hemispheres, and that the following four sulci were also missing:

• MMRR-21-9: left frontomarginal
• NKI-RS-22-2: left paracentral
• NKI-RS-22-14: left frontomarginal
• OASIS-TRT-20-5: right frontomarginal

[binarybottle]

Jason Tourville wrote (20140723):

I reviewed the labeled surfaces for the 4 examples you listed. My notes on each are below. It's possible that manual edits could improve the detection of the fms in the 1st and 3rd scans listed but there are no errors in terms of the identification of the sulcus. Rather, and this also applies to the missing paracentral suclus, the sulcal morphology, makes alignment of the region boundaries to a sulcal fundus difficult or impossible...

MMRR-21-9: left frontomarginal Segmented fms, label boundary deviates mildly from sulcal fundus of medial segment.

NKI-RS-22-2: left paracentral Absence of typical paracentral sulcus, region boundary does not align with sulcal fundus

NKI-RS-22-14: left frontomarginal fms is shallow and unclear, region boundary does not align well with sulcal fundus

OASIS-TRT-20-5: right frontomarginal fms is clear and label boundary aligns with sulcal funds, not sure why fms was not identified.

[binarybottle]

Thank you, Jason! As we discussed over the phone, it is possible that three of the four boundaries may be indistinct due to anatomical variation, but it is quite possible that some of them (as well as all of the callosal sulci) may reside in shallow areas that are not picked up as folds by mindboggle. I am surprised that only 1 of 202 hemispheres did not have a paracentral sulcus!