Hi, ratios are often used, for a nice paper on this topic see; Article Electrophysiological Frequency Band Ratio Measures Conflate ...

For correlation/covariance etc. I think you run into some problems, namely that just frequencies tell you much less than connectivity or time-frequency analysis. For stuff like resting state phase-locking, phase lag and similar are often used, for task-related acitivty time-frequency analysis via wavelet transforms is most often carried out. It's not wrong, just that there are more useful measures that are much more intuitive to interpret and give you much more information on how the brain works.

I hesitate to reply as i am not upto date on brain & volume conduction models, since i have been inactive; now approaching 30 years. I would recommend starting from an review of the available literature on brain and the volume conductivity and then consider the relationship on surface recordings of EEG.

Hi,

cross-correlation could be of help.

Some of these sources could be of help to you:

Siuly Siuly, Yan Li, Yanchun Zhang (auth.)Health Information Science EEG Signal Analysis and Classification: Techniques and Applications [1 ed.] Springer International Publishing 2016.

Munsif Ali Jatoi, Nidal KamelBrain Source Localization Using EEG Signal Analysis [1 ed.] CRC Press 2018

Thank you Bhogaraju Anand I am still confused whether cross correlation can be done using PSD values of frequency bands since many describe doing cross correlation using time series data. Hope you can provide an insight on this too. Thank you so much!

Both (magnitude-squared) coherence and phase-amplitude coupling are common measures to correlate different frequency bands in the EEG. That said, the relevance of these connectivity measures has recently been called into question. Put simply, connectivity doesn't necessarily imply that any information is actually being transmitted.

Article here: Article A mechanism for inter-areal coherence through communication ...

Way less dense press release about the article: https://www.mpg.de/18424976/coherence-turned-upside-down

Hi,

Yes, it is the Time Series model I was referring to.