Hello Zhongming

1. Freesurfer can do this, though it is not perfect. It is my understanding that the best (at least here at the NMR center) BEM segmentations can be obtained when running special pulse sequences called FLASH5 in the MRI to better visualize the skull boundaries. Matti can probably provide additional details.

2. The BEMs outputted are in the appropriate form for EEG/MEG source modeling.

3. The downsampling is a very serious issue that many people have brought up. If you are looking for a first approximation, you will have no problem at all. The freesurfer surfaces can simply be downsampled. However, the spacing between vertices is not uniform. Therefore, downsampling a non-uniformly spaced surface will result in another similarly non-uniformly spaced surface. If you downsample by a large number, you will become acutely aware of the problem. Conversely, if you downsample by a small number, the surface will still look very much like a surface, so perhaps you won't mind.

Another possible solution to this problem, which is still not correct but may be closer, would be to establish a set of points that are uniformly distributed around a sphere, and then pick those points from the cortical surface in spherical coordinates. This solution will suffer from distortions inherent in the spherical morph.

I have been in touch with the makers of SUMA, who are considering developing an option to prepare correctly down-sampled surface vertices based on surfaces Euclidean distances, and our own Doug Greve (also CCed) has also be considering options within the freesurfer environment.

If you consider this issue important enough to solve, and are able to solve it, I encourage you to share it with others, because I have found that a number of people would like the correct solution available to them (myself included).

Daniel

On Tue, 9 Aug 2005, Zhongming Liu wrote:

...

I am a new user of Freesurfer. My questions are related to the head modeling and MEG/EEG source modeling.

1. Can freesurfer generate the 3-shell boundary element model

(BEM): 3 triangulated interfaces of air-scalp, scalp-skull, skull-brain? 2. Is it possible to generate a closed and folded cortical surface after the segmentation on MR images? Freesurfer seems to output two splitted hemispheres of cortex. I am particularly concerning this, because I will use the cortical surface to constrain the MEG/EEG source location and direction. 3. Is it possible to evenly (or nearly evenly) down-sample the vertices of a fine cortical surface into around 6,000 points, onto which the elementary dipole will be placed? Freesurfer generates the cortical surface with a huge number of triangles, which makes it impossible for MEG-based source imaging.

Looking forward to your reply.

Best regards,

Zhongming Liu

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer