Authors: Vinayak Swarnkar; Udantha R. Abeyratne
Addresses: School of Information Technology and Electrical Engineering, The University of Queensland, 78-GP South, Staff House Road, QLD 4072, Australia ' School of Information Technology and Electrical Engineering, The University of Queensland, 78-GP South, Staff House Road, QLD 4072, Australia
Abstract: Estimation of macro-sleep-architecture (MSA) is a critical process in assessing several sleep disorders such as obstructive sleep apnoea, periodic leg movement disorder, upper-airway resistance syndrome, etc. MSA is defined as classification of sleep into three major states: state wake, state REM and state NREM. Existing methods of MSA analysis use six channels of electrophysiological signals (EEG, EOG and EMG). They depend on the manual scoring of overnight data records using the R&K criteria (1968), developed for visual analysis of signals based on morphological features. Manual scoring is cumbersome, subjective and not suitable for portable devices used for community screening of sleep disorders. To address this issue, we propose a fully automated technology for MSA estimation based on a single channel of EEG data. The proposed technology was compared, on a clinical database of 47 patients, with that of an expert human scorer. The average agreement between the human and the proposed technology was found to be 76 ± 7.5% (kappa = 0.51 ± 0.14). The proposed method estimates MSA using simplified instrumentation making it possible to extend EEG/MSA to portable systems as well; method uses low-computation-load bispectrum techniques independent of R&K criteria (1968) making real-time automated analysis a reality.
Keywords: macro-sleep-architecture; MSA estimation; electroencephalograms; EEG; bispectrum analysis; sleep disorders.
International Journal of Medical Engineering and Informatics, 2014 Vol.6 No.1, pp.43 - 64
Available online: 26 Dec 2013 *Full-text access for editors Access for subscribers Purchase this article Comment on this article