Zscore vs Scale with the mean for epileptic discharges power

Hi Mike,

I was wondering whether you have any thoughts on this attached image. These are discharges which I want to compare for power across epilepsy subjects. However I am getting weird results when I baseline normalize with the mean (10*log10(x/u) but the time series on inspection looks ok with z-transform. I tried this with different subjects and its the same - always getting weird visual results with the division normalization formulaes (all tried in brainstorm). The baseline segment I am choosing is artefact and spike free from 500 - 2000 ms before the discharge onset

. Is there any reason you would think so?


Hi Mubeen. I’m not sure what’s going on, but clearly the middle plot looks a bit weird. There could be a bug in the code somewhere? Or just stick with the zscore method.

Thanks Mike, yes currently sticking with z-transforms instead and it is indeed weird.