Coherence NxN configuration

Hi Guys,

In one study, I use 1-minute recordings with participants in a resting state. After treating possible artifacts, I divide the recording into 2-second windows. Then, I apply the "Coherence NxN [2021]" process to the generated window series, using the "Magnitude-squared Coherence" option.

I configure the remaining parameters based on reference studies. However, I have doubts about the process output. Should I select the "Save individual results..." option, which will generate multiple files to later calculate the average, or the "Average cross-spectra..." option, which will generate a single file?
Another point, is it possible to calculate the average coherence values for a frequency range? For example, Theta (4 - 7Hz)... When exporting the file, it asks me to choose a specific frequency (e.g., 5.08Hz). Is it possible to obtain average values for the entire range?

I appreciate your guidance in advance,

  1. "Save individual results..." This option will compute one estimation of MSC for each epoch, using the data only for that epoch, using N windows (N depend on the window length and percentage of overlap). This option can be used to evaluate any change in coherence across epochs.

  2. "Average cross-spectra..." This option will compute one estimation of MSC for all the epochs, using the data from all the epochs, using N windows × number-of-epochs, thus it is a better estimation of MSC.

:bulb: The average of MSC results from Option1 are not the same as the MSC obtained with Option 2 as MSC is not linear.

The Coherence NxN [2021] processes uses the Fourier transform to compute the PSD and cross-spectrum, to later compute MSC. Thus, MSC has the same resolution than FT. To obtain MSC in a band range, a different approach can be take, by filtering data and using Hilbert transform to compute the T-F representation and compute the PSD and cross-spectrum from that T-F representation. More information on these differences here: Lagged coherence - #3 by Raymundo.Cassani

In a not-too-far update, the coherence process GUI will be updated to include both (Fourier transform and Hilbert (or wavelet) transform) approaches

Yes, with the process Frequency > Group in time or frequency bands

1 Like

Thank you for your help, Mr. Raymundo.

So the process mentioned above, splitting a 1-minute file into 2-second blocks, applying MSC Coherence, saving it as 'Average cross-spectrum,' and then averaging the results in frequency bands, is a usual process. Can I use it for all participants and statistical analysis, right?

Thank you again!

Another question: The parameter 'Overlap for PSD estimation' set to 50% is sufficient for a recording with 19 channels and dozens of 2-second windows? Some articles indicate 75% as the value.

Tanks again.