Oscillatory events in the human sleep EEG - detection and properties

A new algorithm for the detection of oscillatory events in the EEG is presented. By estimating autoregressive (AR) models on short segments the EEG is described as a superposition of harmonic oscillators with damping and frequencies varying in time. Oscillatory events are detected, whenever the damping of one or more frequencies falls below a predefined threshold. The algorithm works well for the detection of sleep spindles and in addition identifies delta and alpha waves. The distribution of the occurrence of oscillatory events shows a similar pattern with maxima around 3-4 s for most classes of events except for slow delta waves (< 2 Hz).