Hi Nicky, I think I see what you mean back there. But I can think of only two things that would cause it:
* Cropping the hemisphere too closely to the midline, to where the anatomical volume is getting shaved off before segmentation even begins. This is under your control, so I assume you are not doing that. * The gray matter intensity drops off in the affected cortex to the point where it thresholds out. Based on my understanding of the segmentation algorithm, which hasn't changed for years, there is no cut being applied here by the algorithm. The user crops to hemispheres, so this is usually sufficient, along with the CSF between the two hemispheres, for the hemispheres to separate cleanly. And usually occipital petalia are not shorn off. If you are experiencing version differences, then two things come to mind as possible causes (though these are by no means exhaustive): * spatial normalization causes more torquing * intensities are distributed differently, possibly causing the thresholding issue mentioned above I am not sufficiently familiar with SPM5 vs SPM8 diffs to venture a guess. But it might help to sort out whether these subjects are being spatially normalized before input. If so, then have you tried segmenting the same subjects without normalization (though they'll need to be at least remotely AC-PC oriented for segmentation to work properly). I'm sure it is possible to segment in AC-PC space, then warp nonlinearly to MNI (i.e., apply the warp to the surface coordinates to get something that aligns nicely with your MNI volume). If you're not spatially normalizing first, then look for steps in your SPM pre-processing stream that could affect thresholding. Donna > Donna, > > We apologize for the possible confusion. With 'cuts' we meant that part of > the hemisphere seems cut off after segmentation (see image, top part of > posterior section). In some instances, and often in the more posterior > parts of the hemispheres, a sharp cut-off occurs, so not all grey matter > of a hemisphere seems to be segmented, or part of the segmented > hemisphere. > > In general, we experience less clean segmentation results than before (SPM > 5 and Caret 5.5), leading to more corrections to the segmentation process > and more time dedicated to reconstructing the hemispheres. Additionally, > as mentioned above, more problems in the posterior part of the hemisphere, > leading to difficulties in identifying the Calcarine cut and the > subsequent flattening process. > > Nicky > > -----Original Message----- > From: caret-users-boun...@brainvis.wustl.edu > [mailto:caret-users-boun...@brainvis.wustl.edu] On Behalf Of Donna Dierker > Sent: woensdag 23 april 2014 16:17 PM > To: Caret, SureFit, and SuMS software users > Cc: Hans Op de Beeck; Elfi Goesaert > Subject: Re: [caret-users] Segmentation issue > > Hi Nicky, > > I am a little confused by the "wong cuts" comment. Are these partial or > full hemispheres? Generally, the only cuts in the segmentation process > are ones applied directly by the user during patching. There are also > cuts applied during flattening, but that is a downstream step from > segmentation. > > Some screen captures of the anatomical volume being segmented with > results showing the cuts might be helpful. > > Donna > > > On Apr 23, 2014, at 5:09 AM, Nicky Daniels <nicky.dani...@ppw.kuleuven.be> > wrote: > >> Dear Caret users, >> >> We are experiencing some problems with the segmentation step in Caret. >> The segmentation frequently has erroneous results (for example, wrong >> cuts resulting in parts of the brain being cut off or difficulties in >> clearing up the Calcarine Sulcus). >> >> We seem to have this problem since we switched from SPM5 to SPM8. The >> segmentation step of Caret used to be very convenient when we used SPM5 >> and Caret 5.51 together (OS: windows XP). But since we started using >> SPM8 and Caret 5.65 (OS: windows 7) we are experiencing these problems >> with the segmentation step. >> >> Did anyone of the Caret users experienced these problems as well and >> could anyone help us out? >> >> Thank you very much in advance. >> >> >> >> Nicky >> >> >> _______________________________________________ >> caret-users mailing list >> caret-users@brainvis.wustl.edu >> http://brainvis.wustl.edu/mailman/listinfo/caret-users > > > _______________________________________________ > caret-users mailing list > caret-users@brainvis.wustl.edu > http://brainvis.wustl.edu/mailman/listinfo/caret-users > _______________________________________________ > caret-users mailing list > caret-users@brainvis.wustl.edu > http://brainvis.wustl.edu/mailman/listinfo/caret-users > _______________________________________________ caret-users mailing list caret-users@brainvis.wustl.edu http://brainvis.wustl.edu/mailman/listinfo/caret-users
