Diagnosis of Sleep Apnea Based on Fractal Variability Analysis of EEG Signals in Patients with Obstructive Sleep Apnea

Sleep apnea is a common sleep disorder which is characterized by pauses or instances of infrequent breathing during sleep. Several efforts have been made to detect breathing pauses using HRV (Heart Rate Variability) and EEG (Electroencephalography) signals. As physiological rhythms have chaotic patterns with nonlinear behavior, nonlinear analysis of these characteristics in healthy and diseased conditions provides important information about the pathophysiology of diseases. The aim of this paper is to diagnose sleep apnea using Detrended Fluctuation Analysis (DFA) of EEG signals. After noise and artifact elimination, we separated healthy sleep periods from apnea and calculated DFA for each of them. They were significantly different in stages 1 and 2 of non-REM sleep as well as REM sleep (P<0.05). Since fractal alterations of EEG happen before breathing pauses, apnea is predictable with DFA analysis of EEG signals before it actually happens.