A Global Scale Factor in Brain Topography

Abstract

In this paper statistical resampling techniques are used to support the presence of an equiangular first principal component in EEG log spectra and therefore the existence of a global scale factor in EEG recordings. the Log transformed spectra from a normal sample (n = 211, age 5-97) were analyzed. To reach this conclusion a method for estimating the scale factor is introduced. It is also shown that this factor remains constant in each individual for all derivations and frequencies and across functional states. The contribution of this scale factor to the overall variance of the normal EEG reaches 42% of the total variance of age corrected data. Part of the variance of the scale factor exhibits age dependency. Scale factor correction of EEG spectra data improves the diagnostic accuracy for detecting pathological EEG spectra from an almost random level (Area under the ROC curve = .6) to .84.