search

mne-tools / mne-python

MNE: Magnetoencephalography (MEG) and Electroencephalography (EEG) in Python
https://mne.tools
BSD 3-Clause "New" or "Revised" License
2.66k stars 1.31k forks source link

Temporal connectivity metrics #5802

Open brovelli opened 5 years ago

brovelli commented 5 years ago

Dear all,

at the lab, we are developing a set of functions for directed functional connectivity analysis in the time-domain. These measures are complementary to the frequency-domain metrics, which are already present in mne.connectivity.spectral_connectivity

The list of metrics will not be exhaustive, but we would like to have simple measures to be accessible from MNE. These may include covariance-based Granger causality, Transfer Entropy.

Has anyone already started a similar project in MNE or digging into this issue?

What is your feeling about a similar project? Would a mne.connectivity.temporal_connectivity function be worth the effort?

Thanks for your comments and help

Andrea

agramfort commented 5 years ago

I am not aware of such an effort.

I would recommend to start from a separate project to allow to iterate quickly.

You would expose functions that take mne objects as inputs?

cbrnr commented 5 years ago

We have developed https://github.com/scot-dev/scot, which computes (source-based) connectivity measures based on VAR modeling. Are you referring to such an approach (I'm not sure what you mean by time-domain)? Ideally, all functionality of Scot should be integrated into MNE, but so far I haven't had the time for that.

brovelli commented 5 years ago

@cbrnr : Thanks for the link, we'll look into it. By time-domain FC measures I mean those do not have a frequency-specific decomposition such as Coherence. I normally use the notation described in Bastos et al. 2016

@agramfort : well, the long-term idea would be to make this code compatible with mne objects. We'll probably start off from a separate project.

Thanks again and do not hesitate to suggest coding strategies etc.