BIO-SIGNAL BASED AUTOMATIC SLEEP STAGE CLASSIFICATION SYSTEM CAPABLE OF CONVENIENTLY AND ACCURATELY ESTIMATING A SLEEP STAGE BY ARRANGING A SUBSET WITH AN OPTIMIZED BIO-SIGNAL COMBINATION

摘要

PURPOSE: A bio-signal based automatic sleep stage classification system is provided to reduce the number of measurement signals, thereby conveniently estimating a sleep stage.;CONSTITUTION: A bio-signal detection part(100) measures various bio-signals from each part of a human body using a 7-channel amplifier. The bio-signal detection part comprises two-channel Electroencephalogram(EEG) sensors(110,120), an electrooculogram(EOG) sensor(130), an electromyogram(EMG) sensor(140), a photoplethysmography(PPG) sensor(150), a galvanic skin response(GSR) sensor(160), and a skin temperature(SKT) sensor(170). A signal process and characteristic extraction part(200) extract a feature point which displays characteristics of each bio-signal by processing various bio-signals. A subset generation part(300) arranges a plurality of subsets required for sleep stage classification by mixing the bio-signals with different conditions based on the feature point with respect to each bio-signal. A performance evaluation part(400) analyzes a performance evaluation result of a subset combination according to each sleep stage.;COPYRIGHT KIPO 2012