Compute noise covariance for different sessions

Dear MNE community,

Is it possible to estimate the source of brain activity of a contrast between two different sessions?

If I were to estimate src for data originated in just one session, I know I’d have to compute the noise covariance and the inverse operator. However, the function ‘mne.compute_covariance’ only accepts epochs as input and a contrast between two sessions would always be evoked (I guess).

Can I compute the noise covariance using the epochs of just one of the sessions and then apply it to the inverse operator? Or should I compute the noise covariance for each session separately and then somehow subtract one from another?

epochs_session_1 = epochs_session_1['Hit']

epochs_session _2 = epochs_session_2['Hit_2']

contrast = mne.combine_evoked(epochs_session _1.average(), 
                              epochs_session _2.average() , 
                              weights=[1, -1])
noise_cov = mne.compute_covariance(#What should I pass here as input?, 
                                   method=['shrunk', 'empirical'],

If I was not clear, I would be happy to explain further details.

Best Regards,


MNE-Python version: 0.22.0
operating system: Ubuntu 20.04.2 LTS

if it’s MEG maybe using empty room for the noise cov?

if it’s EEG I would compute the nosie cov on the baseline on the concatenated
epochs from the 2 sessions


But if it’s two sessions, they’d usually have two empty-room recordings as well. How would one handle this?