RESULTS
The proposed method was implemented in MATLAB. This work
discriminates between the awake stage and sleep stage 1 from the
EEG signals in the PhysioNet database. A sample waveform of the
EEG signal from the dataset is shown in the Figure 2.
To evaluate the performance of our work, accuracy (Acc), sensitivity
(Se) and specificity (Sp) are calculated and shown in Table I.
Efficient Sleep Stage Classification Based on EEG Signals
Khald A. I. Aboalayon Helen T. Ocbagabir Miad Faezipour
Department of Computer Science and Engineering The Mathworks, Inc. Computer Science & Engineering and Biomedical Engineering
University of Bridgeport, CT 06604,USA Natick, MA, 01760, USA University of Bridgeport, CT 06604,USA
kaboalay@my.bridgeport.edu Helen.Ocbagabir@mathworks.com mfaezipo@bridgeport.edu
ABSTRACT
Currently, sleep disorders
are considered as one of the
major human life issues.
There are several stable
physiological stages that the
human brain goes through
during sleep. In this work,
Butterworth band-pass
filters are designed to filter
and decompose the
Electroencephalogram
signal (EEG) into five sub-
bands δ, Ɵ, α, β and γ. In
addition, various
discriminating features
including energy, standard
deviation, entropy are
computed and extracted
from above frequency sub-
bands. The features are
then fed to a supervised
learning classifier; support
vector machine (SVM) to be
able to recognize the sleep
stages and identify if the
acquired signal is
corresponding to awake or
stage 1. The experimental
results on a variety of
subjects verify the high
classification accuracy of
the proposed work with
92.5 %.
SIGNIFICANCE
The Electroencephalogram
(EEG) signal is the most
important signal in sleep
stage classification [1]. It
can be calculated by placing
dozens of electrodes at
various sites on the head of
a subject. According to [2]
human sleep is divided into
two stages, Rapid Eye
Movement (REM) sleep and
Non-REM (NREM) sleep.
NREM sleep is further
separated into 4 stages in
which the eyes are usually
closed and many nervous
centers are inactive, so the
brain awareness completely
or partially loses
consciousness and becomes
a less complex system.
Nowadays, many
biomedical signals such as
EEG, ECG, EMG, and EOG
offer useful details for
clinical setups that are used
in identifying sleep
disorders [3].
PROPOSED METHOD
The objective of this work is to propose an efficient technique that
could easily be implemented in hardware to differentiate sleep
stages which will assess physicians to identify certain patterns such
as detecting fatigue, drowsiness, and/or various sleep disorders such
as sleep apnea. The flow chart of the methodology is shown in
Figure 1. First, EEG dataset inputs were obtained from PhysioNet [5]
that were acquired and described by scientists for analysis and
diagnosis of sleep stages. Infinite impulse Response (IIR)
Butterworth band-pass filters are used to decompose the EEG signal
into five different EEG frequency bands. Feasible set of features
including energy, standard deviation and entropy are then
computed and extracted from each δ, Ө, α, β and γ sub-band.
Finally, the features are trained and tested using SVM algorithm to
be able to recognize the sleep stages state.
Classification of EEG signal
Support Vector Machine
Statistical Features Extraction
Filtering and Decomposition into
five EEG sub-bands
Input EEG signal
Figure 1. EEG classification methodology
DISCUSSION and
CONCLUSION
In this paper, we presented
an efficient technique that
could be implemented in
hardware to differentiate
sleep stages which will
assess physicians in the
diagnosis and treatment of
related sleep disorders. IIR
Butterworth band-pass
filters are used to filter and
decompose the obtained
EEG signal from PhysioNet
into five sub-bands δ, Ɵ, α,
β and γ. These bands are
used to predict changes in
brain disorder state [4].
Then, the set of features
including energy, entropy
and standard deviation is
computed for each sub-
band. Linear kernel function
in SVM classifier was used
to train and test using the
extracted features to
classify/detect the sleep
stage. In summary, the key
novelty of this work is to
identify the sleep stages
from a publicly available
EEG signal dataset by using
a feasible set of features,
easily implementable filters
in any microcontroller
device, and an efficient
classification method.
Category
No. of
trained
signals
No. of
tested
signals
Correctly
detected
Acc. % Se. % Sp. %
160 40 37 92.5 85 100
Table I: Performance Result
Figure 2. Sample EEG Signal (a)
awake and (b) Stage 1
Awake EEG signal
Stage1 EEG signal
REFERENCES
[1] E. Estrada, H. Nazeran, P. Nava, K.
Behbehani, J. Burk, and E. Lucas, "EEG
feature extraction for classification of
sleep stages," 26
th
Annual International
Conference of the IEEE Engineering in
Medicine and Biology Society, vol. 1, pp.
196-199, Sept. 2004.
[2] M. Vatankhah, M-R Akbarzadeh-T,
and A. Moghimi, "An intelligent system
for diagnosing sleep stages usingwavelet
coefficients," Neural Networks (IJCNN),
The 2010 International Joint Conference
on , vol., no., pp.1,5, 18-23 July 2010.
[3] F. Ebrahimi, M. Mikaeili, E. Estrada,
and H. Nazeran, "Automatic sleep stage
classification based on EEG signals by
using neural networks and wavelet
packet coefficients," 30
th
Annual
International Conference of the IEEE
Engineering in Medicine and Biology
Society, pp. 1151,1154, Aug. 2008.
[4] L. Chen, E. Zhao, D. Wang, Z. Han, S.
Zhang, and C. Xu, "Feature extraction of
EEG signals from epilepsy patients based
on Gabor Transform and EMD
Decomposition," 6
th
International
Conference on Natural Computation
(ICNC), vol. 3, pp. 1243,1247, Aug. 2010.
[5] PhysioNet, www.physionet.org.
