Temporal dynamics analysis for the invasive EEG

Hi all,
here, I brought a figure from a journal paper, time course analysis of normalized power values.
I am trying to split my recording as it looks with 100ms time bins, and I asked one of my colleagues about HOW and he told me to divide the time bin by peak power value.
In this journal, it is saying it is divided by the average power value but why is he suggesting me to divide it by the peak value? He told me something but I couldn’t get it…
What am I missing?

Thanks!!! :slightly_smiling_face:

I think dividing by the average value is safer. Peak values can be useful, but are also more sensitive to noise. Averages are less sensitive to noise.

1 Like

Right? Thanks!
and, do you think this kind of analysis would work for very detailed time course analysis?
Since I am using ECoG data, I’d like to show this method has a really good time resolution that can even show you the transition in 50ms, 100ms scale.
That’s why I’ve been thinking of having certain length time bin and divide by average power during certain period…

Thanks again!!!

If you want to highlight the temporal precision of the ECoG data, then why bin it in the first place? Binning is basically reducing the temporal precision. If your data were originally recorded at 1000 Hz and you bin to 100 ms, then you’re effectively downsampling the results to 10 Hz.

1 Like

ah… you’re right…
It’s probably better to present normalized power in the time course plot then.
Thanks again for your comment!