MNE version 1.1
Operating system: windows 10
For a research project, we would like to simulate EEG data (not based on an existing dataset) with and without bridging between electrodes (i.e., when two or more electrodes have nearly identical signals) using MNE python, so that we can compare how this bridging influences the following preprocessing steps.
As we have no experience with simulating EEG data, we were wondering if somebody has experience with simulating EEG data in MNE and could help us out with this?
All the best
have you seen our simulation tutorials Simulation — MNE 1.0.3 documentation and examples Data Simulation — MNE 1.0.3 documentation? If you can’t figure it out from those, then let us know what is missing / what you still don’t know how to do, so they can be improved!
Thank you for sharing these helpful links! I will see if I can manage to do it with these tutorials; otherwise, I will let you know.
@larsoner: As requested, here are my follow-up questions:
This simulated data forms the base of the project, so it’s really important that this happens accurately. Therefore, it would be really helpful if somebody with experience would be able to do this (and would be involved in the project).
The next step would be to compare how the bridging affects other preprocessing steps (like artifact removal procedures). For example, we found that the artifact subspace reconstruction did mostly not detect the bridged electrodes in an existing dataset (this was just a school project), but it would be interesting to expand and see if some existing procedures do detect the bridged electrodes (e.g., ICA). If not, it’s a good argument that it would potentially add value to adding a bridging check to the current existing preprocessing pipelines, given that there’s no gold standard yet.
This idea was inspired by the paper of Alschuler and colleagues. These authors used weighted averaging to simulate bridging in Matlab. In MNE, you also have a command to check for bridging and I can use this on existing raw data, but for this project, we would want to use simulated data (one dataset without bridged electrodes, and one dataset that is exactly the same but contains a certain amount of bridged electrodes) (this goes beyond my capacity).