Normalizing EEG Power

[Friedrich Vogt <Friedrich.Vogt@SBG.AC.AT>]

At 10:53 23.05.00 -0500, you wrote:
>Hi all,
>
>I am trying to find a way to normalize the EEG data I acquired with
>different subjects on the same experimental conditions.
>What are the most reliable methods known to avoid intoducing artifacts?
>
>Thanks,
>
>Virginie
>

Hi Virginie

In a study we tested the hypothesis that alpha band power is correlated
with memory performance. And that was the way we prosessed the EEG data
over different experimental conditions:
Artifact-free recordings were segmented in consecutive epochs of 4 secs for
each subject, experimental condition, and each lead. Power spectra were
then calculated for each epoch and averaged. Frequency resolution was 0,25
Hz.
To eliminate interindividual variance in absolute EEG power, the spectra
were normalized by expressing each power estimate for each frequency step
(of ,25Hz) as a percentage of total power (mean of all power estimates) in
a frequency window of 4Hz-16HZ. You can decide how broad You want the
window. This power estimates  are termed normalized percent power. Power
spectra were normalized separatly for each subject, experimental condition,
and lead. Simple to do. Reliable 'cause transparent.

See also: Vogt F., Klimesch W., Doppelmayr M. (1998). High-frequency
components in the Alpha Band and Memory Performance. Journal of Clinical
Neurophysiology. 15(2), 167-172

Servus and greetings

Friedrich Vogt