connectivity measures for a single trial

đź’¬ Support & Discussions
1

Hello all,

I am currently working on a project involving functional connectivity metrics. My understanding is that these metrics, PLV for example, are averaged across multiple trials in response to a specific stimulus. However, I am seeking clarification on how “mne_connectivity.spectral.time” operates when applied to a single trial. Does it perform the calculation once for the individual trial, eliminating the 1/N_trial term? Or, does it employ a sliding window strategy across the time segment of my trial, generating multiple values that are then averaged?

For context, I am working to extract connectivity metrics for single trials that correpond to different motor imagery tasks for classification purposes. So,“mne_connectivity.spectral.time” appears to be suitable for my work, but I would like a better understanding of the underlaying theory.

Much appreciated,

Mohammad

2 / 2

2

Hi Mohammad,

The spectral_connectivity_time function estimates the cross-spectral density at each time point in your single trial and then computes the PLV (or any other metric) using a sum over the time points. More details are given in the Notes section of the documentation. It is possible to also optionally average over epochs.

The spectral_connectivity_epochs function computes the cross-spectral density for each epoch in your dataset (one number represents the whole epoch) and then computes the connectivity metric using a sum over the epochs.

For your application, you should probably use the spectral_connectivity_time function.

Hope this helps,
Santeri

1 Like