Hello MNE users,
We are trying to compute single trial analysis at the source level with MNE.
We understand that we do need an empty room recording from the same day of the experimental recording to compute the noise covariance matrix. Unfortunately for one set of data half of these recordings are missing.
My question is the following:
Is it better to use an empty room recording of a previous or following day to our actual recording (taking into account that our setting does not change from day to days)?
Or will it be better to compute an average noise covariance matrix for the empty room recordings we do have and use this average for each of our participants?
Moreover what kind of pre-processing this raw data should we apply to?
I read in a previous thread that filtering from 1Hz-40Hz could be a good idea but do we need to maxfilter this raw data before to compute the noise covariance matrix?
Thanks for your answer
Elisabeth