Data Covariance VS Noise Covariance

Hello There,
Assuming we have EEG data, we can computer covariance by right-clicking on the EEG data, Two options are available, Data Covariance and Noise Covariance. I wonder if there is a difference in term of computation if we choose same data points to do computation?
also I am wondering how difference BrainStorm compuation is from cov in MATLAB.

Thank you

Assuming we have EEG data, we can computer covariance by right-clicking on the EEG data, Two options are available, Data Covariance and Noise Covariance. I wonder if there is a difference in term of computation if we choose same data points to do computation?

The noise covariance is used by the minimum norm estimates.
The data covariance is only used for the LCMV beamformer.
https://neuroimage.usc.edu/brainstorm/Tutorials/SourceEstimation#Method

The two are computed exactly in the same way.
We didn't come up with any specific recommendations on how to define the time segments from which to estimate this data covariance (only the brain response of interest vs. all the epochs for instance...) - Please refer to the beamforming literature you want to relate to.

I am wondering how difference BrainStorm compuation is from cov in MATLAB.

You can check directly in the code how the covariance is computed:

1 Like