Hi all,
I am new to this field. I have read the definition of PSD but could not find any interpretation related to brain. My question is:
Thanks.
Hello @bipvan,
The traces show the “power” of the signal (across all time points) for each frequency (or, rather, for a number of frequency bins).
This corresponds to the average energy of the measured signal for the respective frequencies.
It is not clear that this necessarily corresponds to neural signals. If, for example, your sensors picked up line noise, you will see a peak in the PSD plots around the line frequency, but this doesn’t mean that the participant’s brain produced this signal.
The “outlier” traces for subject 075 are likely faulty sensors. You can click on the traces to retrieve their name. You should inspect their signals very carefully and decide whether or not to exclude them from analysis. Same goes for this one (?) trace that oscillates somewhat for subject 076.
Best wishes,
Richard
Thanks for clarifying. This data is from epileptic patients and they are ieeg signals. Does it make any difference?
I’m not sure, as I don’t have experience working with iEEG. @alexrockhill and @adam2392 might have an idea, though!
Nope, unless it is clear that the signals are epileptic related, then most likely you are just analyzing noise. PSD is just a summary way of looking at the data.
Analyzing epileptic iEEG signals is a fine balance between qualitative and quantitative analysis that requires a tight collaboration with clinical folks who actually look at the data and can tell you whether something is relevant or not. Or a good understanding of what typical neural signals “look like”.
I would just add that there is some work that shows the 1/f negative slope is likely due to the shape of E/IPSPs Inferring synaptic excitation/inhibition balance from field potentials - PubMed
