Problems segmenting MRI image from "MEGSIM"

[martin]

Hi,

I'm trying to use the simulated MEG data provided on the MEGSIM-Portal (http://portal.mind.unm.edu/megsim/) for evaluation purposed. To construct a forward model I'm trying to extract the cortical surface from the MRI data provided.

For this, I converted the DICOM files to a NIfTI file, imported it into BrainVISA, and tried to use the "T1 Pipeline 2007" to perform the segmentation. Unfortunately, the segmentation does not work, i.e., when using the default settings the pipeline stops at "Split Brain Mask" with the following error:

error1.png (39.17 KiB) Viewed 5439 times

I assume this error occurs because the brain mask is not extracted corretly; it looks like this:

mask1.png (57.82 KiB) Viewed 5439 times

mask2.png (57.02 KiB) Viewed 5439 times

The rest of the mask is empty. I assume this problem is related to the histogram analysis. When I change "Vip Histogram analysis" to "Histogram analysis" I get a mask which is too large, i.e., it covers part of the skull.

Any tips on how to fix these problems, which I assume are caused by a problem with the histogram analysis, are highly appreciated. I used BrainVISA before and so far the segmentation always worked without much tweaking.

Best,

Martin

PS: I tried both BrainVISA V 3.2 and 4.0, both have the same problem

[riviere]

Hi,
We cannot really see the brain on the slices you show here, so we cannot see the contrasts in the brain and get an idea of what caused the problem. However it is generally linked to a failure of the bias correction, before the histogram analysis.
Please search the forum for classical problems, for instance:
http://brainvisa.info/forum/viewtopic.p ... 446&p=5268
http://brainvisa.info/forum/viewtopic.p ... 443&p=5261
http://brainvisa.info/forum/viewtopic.p ... t=22&p=127
etc.

Denis

[Jean-Francois Mangin]

Maybe the histogram analysis failed, you should try to modify one of the parameter
for histogram sampling.