How do you compare two power spectra?


I want to compare the power spectra of two electrodes that I down-sampled at 1kHz and low-pass filtered at 49 Hz (until I learn how to do a proper 50 Hz-band filter).

I was wondering if it is correct to compute the absolute delta power (i.e., the area-under-the-curve (AUC of the delta band (0.5 Hz-5Hz) or better compare it as a percentage of the total power of the signal until 20Hz.

The relative numbers (% of the total band) are pretty similar, however, I find small differences in the absolute number, and I would like to compare the same electrodes in different samples (like these bellow).

I would appreciate a comment here.


Hi Jose. As long as the different electrodes are in the same scale (e.g., same reference electrode, same data processing pipeline), then it’s fine to compare the absolute voltage values.

Using the rest of the spectrum to normalize the activity is also possible. But you need to be careful with this approach (as with any normalization), because energy outside the delta band will contribute to the “delta-band” power estimate. This method is also influenced by possible electrode differences in 1/f distributions.

1 Like