CSEIT2062128 | Accepted : 20 April 2020 | Published : 29 April 2020 | March-April-2020 [ 6 (2) : 504-508]
International Journal of Scientific Research in Computer Science, Engineering and Information Technology
© 2020 IJSRCSEIT | Volume 6 | Issue 2 | ISSN : 2456-3307
DOI : https://doi.org/10.32628/IJSRCSEIT
504
An Effective Approach for Sleep Stage Classification Based on PSG Recordings
Mayuri A. Rakhonde, Prof. Ravi V. Mante, Dr. Kishor P. Wagh
Master of Technology Scholar, CSE, GCOE, Amravati, Amravati, Maharashtra, India
Assistant Professor, Computer Science & Engineering, GCOE, Amravati Amravati, Maharashtra, India
Assistant Professor, Information Technology, GCOE, Amravati, Amravati, Maharashtra, India
ABSTRACT
A person spend his one-third of life in sleep. So, paying attention on sleep is necessary at present times.
Appropriate scoring of sleep stages is essential part in recognition of particular sleep disorder. Sleep stage
classification is the process of categorizing polysomnographic (PSG) recordings into different classes. PSG
contains EEG, EMG, EOG signals. In proposed methodology, Power spectral density is used to extract power
features of EEG and EMG signals. A machine learning model of Stochastic Gradient Descent algorithm is used
for classifying extracted features into multi-class sleep stages.
Keywords: Polysomnography, Sleep Stage, Stochastic Gradient Descent, Power Spectral Density
I. INTRODUCTION
Sleep is the essential part of life. Required amount of
sleep is necessary for human being. Sleep has
tremendous effect on our physical health, mental
health and quality of life. Sleep analysis is the
important factor for recognition of sleep disorders[1].
The common problem is occurred in people is having
trouble in sleeping. Not getting sufficient sleep leads
to the physical and mental illness. Sleep disorders
includes sleep apnea, insomnia, narcolepsy, etc. And
for identification of sleep disorders sleep analysis is
important[2].
Sleep stage classification is the process of classifying
physiological signals into different stages. Basically
there are two types of sleep stages named as Rapid Eye
Movement(REM) and Non-rapid Eye
Movement(NREM). Sleep is a cyclical process. Each
sleep cycle contains three NREM stages and one REM
stage and it continues[3]. Typically, a sleeper goes
through 4 to 6 sleep cycles each about 90 to 120
minutes. As specified by American Academy of Sleep
Medicine(AASM), sleep stages are classified into five
stages named as Awake, Sleep Stage 1(NREM1), Sleep
Stage 2(NREM2), Sleep Stage 3(NREM3) and REM. In
sleep stage 1, a person is in light sleep and move
towards sleep stage 2. In sleep stage 2, a person is in
deep relaxation state and move towards sleep stage 3.
In sleep stage 3 a person is in deep sleep where no eye
movement exist. In REM sleep stage, the movement of
eyes occurred and brain gets active. In this stage
dreams occurs. And cycle continues[4]. Physiological
information is recorded for sleep analysis which
contains continuous time signals and it is called as
Polysomnography (PSG). Polysomnographic
recordings includes Electroencephalogram (EEG),
electrooculograms (EOG), electromyograms (EMG),
electrocardiograms (ECG), oxygen saturation and
respiration. PSG signals are divided into short period
of time called sleep epoch. Each sleep epoch is of 30-
second. Among all of these signals, EEG signal and
EMG signals has more significance. EEG signals has
different characteristics at different sleep stages in