[ENH, MRG] Add interpolate bridged electrodes function #10587
Conversation
agramfort
left a comment
There was a problem hiding this comment.
Have you seen this done somewhere? I would be curious to see on the BCI data if it helps in any quantitative way. Here it seems smart in theory but I am not yet convinced it’s practically useful
|
I haven't seen it done, it just seems like the next logical step. There isn't really a ground truth to evaluate against but it seems like a better idea to me than discarding the data. |
|
The interpolation is definitely different, here is the standard interpolation difference with this method
@agramfort, do you have any idea on how to test quantitatively? You could compare similarity across the rest of the subjects that didn't have bridging for that pair. Maybe with mutual information or correlation? I could try adding that. The bottom two traces are the difference between the interpolation method without adding virtual channels and with. |
You could pretty easily take data that does not have bridging, and create bridging-like responses in some electrodes (maybe use the mean of both/all bridged electrodes signals and add a tiny bit of Gaussian noise?). Then you could look at which interpolation method actually matched the original/veridical (un-bridged) data better |
|
Thanks for the idea @larsoner, glad I asked, that was much easier and probably more accurate. It does look like adding a virtual channel gives you slightly less variance of the residual (difference between original signal and recovered). |
|
Thx @alexrockhill |
* upstream/main: MRG: Add "highlight" parameter to Evoked.plot() to conveniently highlight time periods (mne-tools#10614) MRG: Allow to pass array of "average" values to plot_evoked_topomap() (mne-tools#10610) fix: snirf length units (mne-tools#10613) minor, doc: fix subplot titles in tutorial (mne-tools#10607) Display averaged time period in Evoked topomap title (mne-tools#10606) MAINT: Fix for pydata-sphinx-theme [skip azp][skip actions][circle deploy] (mne-tools#10605) DOC: report.add_html in tutorial (mne-tools#10603) Broader support of the SNIRF file format and enable reading GowerLab data (mne-tools#10555) MRG: Recommend mamba instead of libmamba for installation (mne-tools#10597) Fix dev documentation warning [skip azp][skip actions] (mne-tools#10599) FIX cmap (mne-tools#10593) [ENH, MRG] Add interpolate bridged electrodes function (mne-tools#10587)
Follow-up to #10571.
I thought it would be nice instead of just dropping bridged electrodes to use them in the interpolation. This just makes a virtual channel halfway between bridged electrodes and then interpolates with the aid of the virtual channel before dropping it. I thought this would be one last nice thing to add.