Hi all,
I am using empty room MEG recordings for the computation of the noise covariance matrix. I believe that the empty room recording needs to prepared for Maxwell filtering using mne.preprocessing.maxwell_filter_prepare_emptyroom, followed by Maxwell filtering using mne.preprocessing.maxwell_filter. My question is, when in the preprocessing of the actual MEG data should the covariance matrix be computed? If you perform the Maxwell filtering at the beginning, the rank of the task MEG data may change (with ICA or SSP), which could give the covariance matrix a larger rank. Should Maxwell filtering happen later so the two instances have the same rank, or can the rank be specified later?
Thanks all!