[FieldTrip] Noise normalization of MNE

[HINDRIKS, RIKKERT]

Dear all,

I'm having trouble interpreting the statistical parametric maps obtained
from MNE source reconstructions using the method described in (Dale, 2000)
. The issue is the following:

For each source location (assume the dipole orientations fixed) the
noise normalized
activity estimate at that location is (approximately) standard normally
distributed under
the null hypothesis of no activity at all locations.

However, I would like to test for activity at a specific location, but
under the hypothesis
of activity at a specific location, the normalized estimate at
that location is only standard normally distributed if the corresponding
row of the resolution matrix contains a single
non-zero entry.

In other words: it seems to me that the statistical parametric map can only
be used to test
for activity at individual locations if the source reconstruction is
unbiased (or equivalently:
if the resolution matrix equals the identity matrix) which is not the case
in general.

It would be very helpful if someone could help me clarifying this.

Kind regards,
Rikkert
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20160614/0511e522/attachment-0001.html>