Dear all, dear Francois,
is it possible to average the root mean square of oscillatory raw signals within brainstorm and to localize the average afterwards?
Marcel
Hi Marcel,
the RMS average operation is non linear; therefore it would violate the basic MEG/EEG forward model which is: M = G * S, where S are the sources, M are the measurements and G is the forward operator conditioned by the physics of MEG/EEG. Computing the RMS average boils down to : Mrms = sqrt(mean(M.^2)) or similar, hence there is no direct dependency of the new set of measures Mrms with source amplitudes.
What you can do though is compute this RMS average at the source level.
Hope this helps
Dear Sylvain,
thank you for your basic lesson in MEG physics ;-)! Is there a way to average and localize oscillatory activity within brainstorm?
Marcel
Hi Marcel,
One possible way to detect oscillatory activity using BST is to use the tools for time-frequency decomposition. Francois has written a great introductory tutorial at http://neuroimage.usc.edu/brainstorm/Tutorials/TutTimefreq. You can compute the TFD at the sensor level or at the source level, using scouts as regions of interest. Most basic processes in BST apply to TFD maps and therefore you’ll be able to average across trials to detect induced oscillatory components in your data.
Let me know if you have more questions.