Hi Chris,
Do you have any other scans on the same individuals? Multi-channel segmentation might help disambiguate these features if the second channel is e.g. a high-resolution T2 scan.
Best wishes,
Jason
________________________________________
From: SPM (Statistical Parametric Mapping) [[log in to unmask]] on behalf of PRESOTTO LUCA [[log in to unmask]]
Sent: 22 November 2017 15:58
To: [log in to unmask]
Subject: [SPM] R: [SPM] Wide diploic space disrupts segmentation
Dear Chris,
I had the exact same issue. Nonetheless, I haven't been able to find a solution for it, despite trying all the tricks I could think of. In the end, I discarded the subjects.
The only solution I can think of is manual masking.
Luca
-----Messaggio originale-----
Da: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] Per conto di Chris Rorden
Inviato: mercoledì 8 novembre 2017 16:59
A: [log in to unmask]
Oggetto: [SPM] Wide diploic space disrupts segmentation
Hello all,
I am working with a clinical dataset that includes 181 older adults. We use SPM12's segmentation and DARTEL. This generally works extremely well. However, 7 individuals have widening of diploic space and for these cases the segmentation mis-identifies the skull's cancellous bone (bone marrow) as being gray matter, yielding poor segmentation and normalization. I recognize the issue is rare and relatively minor, but I still wonder if there is a good solution. I have tried every SPM trick I can think of (lossless cropping of volume, good starting estimate, tissue cleanup, tuned bias regularisation, using old 3-tissue segmentation, using oldNorm to warp brain masks) to no avail. The only hacks I have found that work is to darken the tissue outside the brain about 50% percent based either on laborious manually drawn masks or to generate based on antsBrainExtraction (though the latter also often fails with these individuals). I am interested if anyone else has any thoughts on how they deal with these issues, and whether there are better solutions.
Rispetta l’ambiente: non stampare questa mail se non è necessario.
Respect the environment: if it's not necessary, don't print this mail.
|