 |
 |
Dear Sermin, The "holes" are actually gaps into the surface from outside and are always at the boundary (and hence connected to other outside voxels and not just internal). Therefore there is nothing wrong with the result. The white matter is quite pronounced at the edge of your caudate in this image and the boundary correction is giving you a jagged edge because that is how it appears in the image. However, it may not be optimal so you could try different boundary correction methods with the first_boundary_corr tool (see the documentation). If you look at the image prior to boundary correction (the *_first.nii.gz image) then you'll see that it has done a good job at segmentation, but it is simply the fact that it has a smooth boundary whereas the image shows a jagged boundary with the infiltrating white matter which is causing a problem. As I said, try different boundary correction methods if you are unhappy with this result. All the best, Mark On 24 May 2011, at 16:06, SERMIIIN Dz wrote: > Dear Mark > > I have tried again and have uploaded the nifti images with the reference number: 380594. > Hope that it is better. > Thanks a lot > > Sermin Dzaferov > > > > > > Date: Tue, 24 May 2011 14:19:27 +0100 > > From: [log in to unmask] > > Subject: Re: [FSL] segmentation error > > To: [log in to unmask] > > > > Dear Sermin, > > > > Are you sure that these are not connected in the other > > planes? In 3D if you have slightly jagged boundaries > > then sometimes it looks like there is a hole in one 2D slice, > > but if you look at the other slices you'll see that it is > > actually just part of the boundary. > > > > I was expecting to get the nifti images, rather than > > screenshots, so I can't really say anymore than that. > > I would also look at the non-boundary-corrected image > > and see if it might be the case that the boundary > > correction is not doing well and might be changed. > > > > All the best, > > Mark > > > > > > On 24 May 2011, at 14:06, SERMIIIN Dz wrote: > > > > > Dear Mark > > > > > > Thank you for your reply. For your information we are dealing with children in the age 7-12 years old. I've uploaded 2 examples illustrating my problem. The reference number is 700071. > > > Hope to hear from you soon. > > > Thanks a lot, > > > > > > Sermin Dzaferov > > > Bachelor-student > > > Technical University of Denmark, > > > > > > > > > > > > > > > > Date: Mon, 23 May 2011 18:03:08 +0100 > > > > From: [log in to unmask] > > > > Subject: Re: [FSL] segmentation error > > > > To: [log in to unmask] > > > > > > > > Dear Sermin, > > > > > > > > You should not get holes in structures. > > > > You should also not use FIRST with children younger than 4 years > > > > old, as that is the limit of the age range for the training data used. > > > > If it helps you can upload an example segmentation (and original > > > > image) to: > > > > http://www.fmrib.ox.ac.uk/cgi-bin/upload.cgi > > > > and send us the reference number so that we can see the problem. > > > > > > > > All the best, > > > > Mark > > > > > > > > > > > > On 23 May 2011, at 15:57, SERMIIIN Dz wrote: > > > > > > > > > Hi > > > > > > > > > > My partner and I have segmented some subcortical structures using FSL-FIRST. As we want to check the segmentation in FSL-view, we discover some "unsegmented" areas which appear as a hole. Why does this happen? and how can this be avoided? Our study goes on children and the brain-maturation. Does the program have any problems dealing with a child's brain? I've tried to read the Technical report/thesis by Brian Patenuade but without success. > > > > > > > > > > Thanks in advance. > > > > > > > > > > /Sermin Dzaferov > > > > > Bachelor-student > > > > > Technical University of Denmark, > > > > >