To detect REM and wake stages in sleep, this study generates feature variables from the correlation of two-channel EOG signals and the amplitude of LEOG signal. By using the VQ method to quantize these signals into different codewords and by calculating the number of appearances of these codewords, we are able to establish a feature vector for every epoch of the recorded EOG signals. Via a three-stage process, the personalized classification accuracy for REM and wake sleep stages are about 95% and 86%, respectively. By combining these personalized classifiers to perform REM and wake stages detection for other unseen individuals, the classification accuracy for REM and wake sleep stages, the classification accuracy become 85% and 92%. However, the sensitivity for the wake stage detection is merely 52%.
Identifer | oai:union.ndltd.org:NSYSU/oai:NSYSU:etd-0718108-170721 |
Date | 18 July 2008 |
Creators | Wang, Yen-shi |
Contributors | Jiann-der Lee, Chen-wen Yen, Liang-wen Hang |
Publisher | NSYSU |
Source Sets | NSYSU Electronic Thesis and Dissertation Archive |
Language | Cholon |
Detected Language | English |
Type | text |
Format | application/pdf |
Source | http://etd.lib.nsysu.edu.tw/ETD-db/ETD-search/view_etd?URN=etd-0718108-170721 |
Rights | unrestricted, Copyright information available at source archive |
Page generated in 0.0052 seconds