I have been trying to compare the different source reconstruction methods doing a simple simulation. I have ECoG data and I already created the forward model. I selected only one vertex as the scout (active dipole in the simulation) and used the "Simulate section" process to create a signal on it and to map it to the ECoG sensors adding a 3SNR noise. Then, I tried different methods (MNE - with and without normalization-, LCMV, dipole moleding) to reconstruct the source but always the performance is not good enough. My source is located on the right posterior hemisphere and the reconstruction detects it at the left frontal (Images attached).
I'm not sure where the problem could be? Any suggestions?
I'm not sure how realistic this is to use the forward model computed for the ECOG sensors to run simulations like the one you describe... Two issues:
The forward model for ECoG and sEEG have never been properly validated. They are available in Brainstorm more for testing, and as a way to project the electrodes values on the cortex. You should rather consider this as an undocumented feature...
The forward model is incomplete: for most sources, there is no information available... the forward model for the sources that are more than a centimeter away from the ECoG contacts is quite random. Therefore, the minimum source maps computed with a full cortex can be very random: some combinations of ECoG values might lead to unexpectedly high values away from the recordings sites, and could even mask obvious effects that are correctly recorded.
I'm not sure I would recommend working in this direction...
Thanks for your answer. I'm trying to identify the sources during epileptic seizures and I have ECoG and sEEG; accordingly with the information, it's not reliable to do it using brainstorm? I've done also the Epileptogenicity tutorial, but unfortunately the process didn't work (I'm using the last version of brainstorm)
sEEG and ECoG give you information that is valid only for a few millimeters around the contacts.
Trying to reconstruct the activity of the full brain with a minimum norm solution is not really meaningful.
The most informative displays you can have are the signals themselves.
If you want to have a spatial representation of these signals, prefer simple interpolations on the cortex surface of the MRI volume: https://neuroimage.usc.edu/brainstorm/Tutorials/ECoG#A2D.2F3D_topography
I've done also the Epileptogenicity tutorial, but unfortunately the process didn't work
Have you followed the tutorial using the example dataset provided in the tutorial?
What error do you get?
Okay, thank you Francois. What do you think about the published papers (Kim et al., 2010 NeuroImage; Pascarella et al., 2010 Neuroscience Methods) in which source reconstruction is done using ECoG? I'm trying to rethink the project I have in hands
About the tutorial, yes, I used the example dataset and followed exactly the tutorial. I have this error:
Line 37: fileparts (line37)
Input must be a row vector of characters or string scalar.
Call stack:
fileparts.m at 37
bst_fileparts.m at 55
in_fopen.m at 47
export_data.m at 106
process_epileptogenicity.m>Run at 219
process_epileptogenicity.m at 28
bst_process.m>Run at 231
bst_process.m at 36
panel_process2.m>RunProcess at 150
panel_process2.m at 26
gui_brainstorm.m>CreateWindow/ProcessRun_Callback at 772
bst_call.m at 28
gui.brainstorm.m>@(h,ev)bst_call(@ProcessRun_Callback) at 296
I have no opinion, I have no experience with source estimated from ECoG data, you should rather contact the authors of these articles (@AnnalisaPascarella), or the OpenMEEG team (@Alexandre, @mclerc, @papadop). If they think it is correct to use the forward models computed for ECoG with OpenMEEG for minimum norm imaging in Brainstorm, then OK. But we won't be able to provide any help regarding the weird activations you observe away from your recording sites.
There are probably things you can try with your recordings before going for this advanced electromagnetic modelling methods.
About the tutorial, yes, I used the example dataset and followed exactly the tutorial. I have this error:
[process_epileptogenicity] Epilepsy > Epileptogenicity maps (A=Baseline,B=Seizure)
Line 37: fileparts (line37)
This bug is already fixed. Update Brainstorm and try again.
(Images attached).
