| Automated sleep staging based on biological signal analysis provides an important quantitative tool to assist neurologists and sleep specialists in the diagnosis and monitoring of sleep disorders as well as evaluation of treatment efficacy. A complete visual inspection of the EEG, EOG and EMG recordings acquired during nocturnal polysomnography (NPSG) is time consuming, expensive, and often subjective. In this thesis different feature extraction schemes from EEG, EOG and EMG signals are presented: Amplitude, Relative Spectral Band Energy, Hjorth Parameters, Harmonic Hjorth Parameters, Itakura Distance, and Detrended Fluctuation Analysis (DFA) statistics. Then, the performances of these schemes are compared to select an optimal set of features for specific, sensitive, and accurate neuro-fuzzy classification of sleep stages. Artificial neural networks (ANNs) have the ability to divide the derived feature vectors in a nonlinear fashion and allow detecting sleep stages from a specific combination of these feature sets. |
