Artifact Removal in Physiological Signals Using Singular Value Decomposition

Sleep staging is a laborious and time consuming process performed by sleep specialists in sleep laboratories. Patient's sleeping process is analyzed after this process. These evidences give direction to the treatment process for many diseases that are described as sleep disorders in the literature. Automatic sleep staging systems are computer aided systems designed to make these operations more reliable and faster. The decision making mechanisms of these systems are based on signal processing and classification. For this reason, working with artefact-free signals is of great importance for increasing the classification accuracy. This study aims to eliminate the negative effects of Electrooculography (EOG) artifacts on sleep Electroencephalography (EEG) signals. The proposed artefact elimination process has been performed using the Singular Value Decomposition method and this method is very useful for online sleep staging systems.