EEG simulations

Dear all

I believe this has been posted elsewhere, however I have been searching and always find incomplete instructions on how to do what I need. I would like to be able to simulate:

  1. EEG background like activity
  2. ERPs
  3. Others (epileptic activity, etc)

What I would like is a procedure from scratch. I could do as suggested in http://neuroimage.usc.edu/forums/showthread.php?846-EEG-simulation, however then I would need to define the entire ImageGridAmp matrix in order to simulate different background EEG than the one used to obtain the Noise Covariance matrix and the source activity matrix (if I understood correctly).

Is there a way to use the new simulation features to do 1-3 above? In addition, imagine I want to begin from the very basics by using the default Colin27 anatomy and a 3-shell sperical model, then how do I define a scout? It does not allow me if I have no Noise Covariance matrix…

Any help would be appreciated
Toni

Dear Toni,

There is nothing in Brainstorm readily available for simulating EEG. We do not provide any noise models that would generate physiologically plausible EEG signals.
You can simulate your own signals and use Brainstorm to display your simulated sources, calculate a forward model, multiply the two (EEG = inverse * forward) and review the simulated EEG.

If you want to start from a few scout signals saved in “matrix” files (such as what is created by the processes “Simulate generic”, “Simulate PAC” and “Simulate AR”):

  • Create a new subject, using a default anatomy and channel shared per subject
  • Open the cortex, create a few scouts
  • Right-click on (Common files) > import your channel file
  • Right-click on (Common files) > Compute head model
  • Run one of the Simulate process for this subject, with the Process1 box empty
  • Select the generated file in Process1 and run the process “Simulate > Simulate recordings from scouts”, select a number of scouts that matches the number of signals

If you want to define manually the full 15000 sources, from the previous setup:

  • Right-click on the (Common files) > Noise covariance > No noise modeling
  • Right-click on the simulated EEG recordings > Compute sources > Select [Expert mode] and “Full results”
  • Right-click on the new source file > File > Export to Matlab > ResMat
  • Edit the structure: ResMat.DataFile=’’; ResMat.ImageGridAmp=your data
  • Right-click on the same file > File > Import from Matlab > ResMat.
  • Right-click on the new customized results matrix > Model evaluation > Simulate recordings

Cheers,
Francois

Dear all

You could tell that represents the gridAmp matrix, and what signifcan its rows and columns ?.

I have an array (so far) of 15002x10000. I can infer that 1000 is the simulation time, but 15002 is each of the grid points? where there may be a source ?.

Greetings await your response.

Hello,

The size of the ImageGridAmp matrix is [Nsources x Ntime].
For a better understanding of this files, please follow the 12+3 introduction tutorials, especially the one about source estimation:
http://neuroimage.usc.edu/brainstorm/Tutorials/TutSourceEstimation

Cheers,
Francois