the correspondence between line color and tissue type

💬 Support & Discussions
, ,
1
  • MNE version: 0.24.0
  • operating system: CentOS release 6.5 (Final)

Hi! all,
I used

mne.bem.make_watershed_bem(subject, subjects_dir=subjects_dir, overwrite=False, volume='T1', atlas=False, gcaatlas=False, preflood=None, show=True, copy=False, T1=None, brainmask='ws.mgz', verbose=None)
mne.viz.plot_bem(subject, subjects_dir=subjects_dir, orientation='coronal', slices=None, brain_surfaces='white', src=None, show=True, show_indices=True, mri='T1.mgz', show_orientation=True)

to compute and visualize the bem surface, here is what the result look like:

image
image1837×967 108 KB

since the verbose showed that:

Using surface: /brain/guixue/zhouyu/proj01/mri/freesurfer/recon_all/sub-41/bem/inner_skull.surf
Using surface: /brain/guixue/zhouyu/proj01/mri/freesurfer/recon_all/sub-41/bem/outer_skull.surf
Using surface: /brain/guixue/zhouyu/proj01/mri/freesurfer/recon_all/sub-41/bem/outer_skin.surf
Out[25]: <MNEFigure size 780x731.5 with 12 Axes>

I thought the red contour corresponds to the inner skull, the yellow contour corresponds to the outer skull and the fleshcolor, i.e., the outermost outline corresponds to the outer skin, i.e., the scalp surface. Am I right? I’m not sure about this :thinking:

By the way, did the segmentation looks okay? Is it good enough for the MEG source analysis?

2

I thought the red contour corresponds to the inner skull, the yellow contour corresponds to the outer skull and the fleshcolor, i.e., the outermost outline corresponds to the outer skin, i.e., the scalp surface. Am I right? I’m not sure about this :thinking:

this is correct. You see that the watershed BEM algorithm overestimates the skull thickness in the lower part. MNE
recommends a flash sequence to avoid this model approximation.

By the way, did the segmentation looks okay? Is it good enough for the MEG source analysis?

looks good enough given only a T1 image

Alex

1 Like
3

Okay~maybe I can try both (watershed & flash) for subjects who had T1 and flash images to see how far are the final analysis results obtained from these two methods different. Because after all nearly half of my subjects didn’t have the flash images.
Thank you :blush: :blush: :sparkling_heart: :sparkling_heart:

4

Hi, Alex

I have the same question as YuZhou.

I used the command lines in the Prompt(MNE) window to compute the BEM surfaces of one subject:

export FREESURFER_HOME=/Applications/freesurfer/7.3.2
export SUBJECTS_DIR=/Users/yr/freesurfer_recon 
source $FREESURFER_HOME/SetUpFreeSurfer.sh

mne watershed_bem -s sub01 -d /Users/yr/freesurfer_recon

I used my data and got the figure in which the yellow contour(outer skull) is very close to the orange contour(outer skin) and far from the red contour (inner skull), as indicated in the blue rectangle.

BEM surfaces-my subject
BEM surfaces-my subject1920×1032 123 KB

However, in the Tutorials, the yellow contour is close to the red one, as indicated in the white rectangle.

BEM surfaces-mne sample
BEM surfaces-mne sample1920×937 130 KB

I think my result is wrong because the yellow contour looks incorrect. I’m not sure whether my result is right or wrong. If it is wrong, how can I solve it? How should I modify the $ mne watershed_bem line?

BR,
Ye

5

this is explained by the fact that you use mne watershed_bem and not mne flash_bem as
we do in the doc. So it’s an expected result although admittedly not perfect. We don’t have
a better solution at this point

Alex

6

Oh! I see. The ‘Head model and forward computation’ in the tutorial does not specify which algorithm is used. But now I know, I will try mne flash_bem. Thank you, Alex!