 |
 |
Dear Stephen, On Sun, 18 Jun 2017 02:10:09 +0200, Stephen LARROQUE <[log in to unmask]> wrote: >Also I have a subsidiary practical question: if you know a particular >subject has a difficult segmentation but you can fix it by tweaking the >tissue segmentation parameters, is it ok to do so for only this subject and >do an inter-subjects VBM statistical test with other subjects using >different parameters? If the parameter change will be only necessary for a few subjects I don't see an issue here. However, if you always have systematical changes only for the patient group and not the control group this will cause very likely a bias in your preprocessing and analysis. Best, Christian > >To clarify, I know that usually one should use the same parameters for all >subjects of a study, but here the goal is to correctly segment a subject, >if all subjects are correctly segmented but one, adjusting segmentation >parameters for this subject seems akin to manual segmentation. Would this >practice be technically ok or would it be biased? > >Thanks a lot for your feedback! >Stephen L. > >2017-06-17 17:04 GMT+02:00 Stephen LARROQUE <[log in to unmask]>: > >> Hello everyone, >> >> I am using CAT12 to segment a brain to do a VBM analysis. >> >> The brain is quite damaged, and some areas, particularly the frontal lobe, >> suffer from inhomogenous intensity of the white matter. >> >> However, the CAT12 report indicates that the segmentation could separate >> the white matter tissue in the "GWM" class (I guess meaning it cannot know >> if it's grey matter or white matter?). I posted a picture of the report >> here: >> >> http://imgur.com/a/leLOH >> >> I would like to know if it is possible to force CAT12 to only use the >> tissue classes GM and CGM for the normalization, excluding the GWM class. >> >> Also, what does CGM stands for? Cerebrospinal-fluid mixed with Grey Matter? >> >> Thank you very much in advance, >> Best regards, >> Stephen L. >> >