Does renaming the argument `noise_cov` to `data_cov` in the function make_inverse_operator make sense?

💬 Support & Discussions
1

MNE version: 1.7.0
operating system: Windows 10

Hello,

I was following this tutorial to perform source analysis for my epoched data. I saw we used mne.minimum_norm.make_inverse_operator to construct the inverse operator. One of the key arguments is the noise_cov. I found the word ‘noise’ is a bit confusing, as I saw in the example, the data used to compute the covariance matrix seemed to be the event-related epochs rather than some non-event-related data segments that is usually used for accounting uninterested noises.

For example, in mne.beamformer.make_lcmv, it dinstinguishes between data_cov and noise_cov. Indeed, the original LCMV paper emphasized the importance of supplying a noise covariance. While the function mne.minimum_norm.make_inverse_operator is a basic helper function for many things not limited to minimum norm source localization, would it make sense to rename noise_cov to data_cov to make it less specific and reduce potential confusion?

I might be mistaken about some aspects mentioned above, so I apologize if my understanding of the code dependencies and related methods is incomplete. Ultimately, I want to understand the intention of some variable names to ensure I use it correctly. So any comments on why this is a good or bad idea are very welcome.

Thank you,
ST

2

Hello, the code used in the tutorial is:


noise_cov = mne.compute_covariance(
    epochs, tmax=0.0, method=["shrunk", "empirical"], rank=None, verbose=True
)

The tmax=0 means that only the pre-stimulus period is considered for the covariance calculation. Since this period can usually be assumed not to contain any signal of interest, it does make sense to treat it as “noise” here.

Best wishes,
Richard

1 Like
3

Hi,

You can read this paper https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=34b659eeb98727fc37b7e60c0d652fcb20ccb870 to know more about noise covariance estimation for inverse modeling

Alex

2 Likes
4

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.