[Mne_analysis] MNE vs. beamforming

Ghuman, Avniel (NIH/NIMH) [F] ghumana at mail.nih.gov
Fri Feb 26 17:43:14 EST 2010
Search archives:

Hi everyone,

I am trying to understand the fundamental difference between beamforming and
MNE.  The major thing is that I am trying to understand how they use the
noise covariance differently.  In particular, it seems to me that people who
use beamformers assume that MNE does not use the data itself to create the
inverse solution (for example, the first paragraph on page 2388 of Quraan et
al NeuroImage 2010 says "Non-adaptive spatial filters rely on the forward
solution in their formulation of the inverse problem in a manner completely
independent of the measurement, while adaptive spatial filters rely on both
the forward solution and the measurement. The popular minimum-norm solution
can be formulated as a non-adaptive spatial filter.").  However, clearly we
do use the prestim/noise data to create our noise covariance matrix.  What I
am trying to understand is: Am I missing something fundamental about what
they mean by using the measurement in formulating the inverse solution (if
so, what?)?

Also, I know that often the beamformers use the evoked data itself in their
covariance matrix.  However, it seems like beamformers could, in theory, use
prestim time or empty room noise, as the MNE stream does, for its covariance
matrix.  If one were to employ a beamformer, but use noise to build the
covariance matrix, would the method differ greatly from MNE?  If so, how
exactly?

Many thanks in advance.

Thank you,
Avniel





More information about the Mne_analysis mailing list