[Mne_analysis] Lead field spikes using sample BEM - bug or explainable?

[Taylor Williams]

        External Email - Use Caution        

Thank you. To close this out, using mne.surface._points_outside_surface()
in v0.20, I was able to compare the sensor locations (manually transformed
into MRI coordinates first) to the BEM mesh for the outer skin surface.
About 10% of the sensor locations in the model are inside the outer skin
surface. I'm working on an alternate approach to generating high density
on-scalp sensor locations now.

Taylor

On Thu, Feb 13, 2020 at 6:49 AM Eric Larson <larson.eric.d at gmail.com> wrote:

> The curious thing is that, when I view the co-registration with the
>> 'head-dense' surface shown, it appears as though there are no collisions
>> with the outer skin surface.
>>
>
> The BEM is made from a smoothed, downsampled version of this surface. If
> you use `plot_alignment` and use `surfaces='head'` or
> `surfaces='outer_skin'` it might show you that one instead.
>
>
>> Do you know of a decent way to detect any sensor locations that are
>> inside the outer skin BEM surface to avoid problems like this in the future?
>>
>
> You can use some private functions/classes
> <https://github.com/mne-tools/mne-python/blob/master/mne/surface.py#L546-L590>
> from `mne.surface` for this, with the caveat that the API is not guaranteed
> (subject to change at any time without maintaining backward compatibility).
>
>
>> Does the outer skin surface include things like the ears?
>>
>
> Yes it will but these will be smoothed in some fashion for the BEM.
>
> Eric
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200213/6e995c12/attachment.html