Conference PaperPDF Available

Detection of Obstructive Sleep Apnea Through ECG Signal Features

Conference Paper

Detection of Obstructive Sleep Apnea Through ECG Signal Features

Abstract and Figures

Obstructive sleep apnea (OSA) is a common disorder in which individuals stop breathing during their sleep. Most of sleep apnea cases are currently undiagnosed because of expenses and practicality limitations of overnight polysomnography (PSG) at sleep labs, where an expert human observer is needed to work over night. New techniques for sleep apnea classification are being developed by bioengineers for most comfortable and timely detection. In this paper, an automated classification algorithm is presented which processes short duration epochs of the electrocardiogram (ECG) data. The automated classification algorithm is based on support vector machines (SVM) and has been trained and tested on sleep apnea recordings from subjects with and without OSA. The results show that our automated classification system can recognize epochs of sleep disorders with a high degree of accuracy, approximately 96.5%. Moreover, the system we developed can be used as a basis for future development of a tool for OSA screening.
Content may be subject to copyright.
Detection of Obstructive Sleep Apnea Through
ECG Signal Features
Laiali Almazaydeh, Khaled Elleithy, Miad Faezipour
Department of Computer Science and Engineering
University of Bridgeport
Bridgeport, CT 06604, USA
{lalmazay, elleithy, mfaezipo}@bridgeport.edu
AbstractObstructive sleep apnea (OSA) is a common disorder
in which individuals stop breathing during their sleep. Most of
sleep apnea cases are currently undiagnosed because of expenses
and practicality limitations of overnight polysomnography (PSG)
at sleep labs, where an expert human observer is needed to work
over night. New techniques for sleep apnea classification are
being developed by bioengineers for most comfortable and timely
detection. In this paper, an automated classification algorithm is
presented which processes short duration epochs of the
electrocardiogram (ECG) data. The automated classification
algorithm is based on support vector machines (SVM) and has
been trained and tested on sleep apnea recordings from subjects
with and without OSA. The results show that our automated
classification system can recognize epochs of sleep disorders with
a high degree of accuracy, approximately 96.5%. Moreover, the
system we developed can be used as a basis for future
development of a tool for OSA screening.
Keywords: sleep apnea, PSG, ECG, RR interval, feature
extraction, SVM.
I. INTRODUCTION
A. Background
Over the average lifespan, humans sleep for about 1/3 of
their lives. Feeling terrible after a night without sleep is the
body’s way of reminding us that sleeping is a necessity such as
eating, drinking and breathing. As we sleep, our body repairs
itself. This rejuvenation goes for hormones and muscles as well
as neural responses, like memory. Without sleeping, we simply
do not function as well as we can.
A sleeping disorder is when one cannot sleep, causing the
body to lose function. Just as the body’s benefits of rest can
range from physical to emotional and psychological effects,
lack of sleep can damage the body physically, emotionally and
psychologically. Till date, 84 kinds of sleep disorders have
been discovered, including the most common ones such as
insomnia, sleep apnea, narcolepsy and restless leg syndrome
[1].
Sleep apnea (SA) is the instance when one either has pauses
of breathing in their sleep, or has very low breath while asleep.
This pause in breathing is known as an apnea, and can range in
frequency and duration. The lack of breathing during sleep is
known as a hypopnea [2].
Sleep apnea is classified into two different types. The first
type is Obstructive Sleep Apnea (OSA), which is more
common, occurring in 2% to 4% of middle-aged adults and 1%
to 3% of preschool children [3], and is generally caused by a
collapse of the upper respiratory airway. The second one is
Central Sleep Apnea (CSA), which is caused by an absent or
inhibited respiratory drive. Most cases of CSA are mixed,
meaning that it is often along with OSA cases, and the CSA
only cases appear exceedingly rarely [4].
Like all sleeping disorders, symptoms of sleep apnea do not
occur just during the night. Daytime symptoms can range from
excessive sleepiness, impaired concentration, depression, early
morning headaches, memory loss and irritability [3].
During the night, symptoms can include nocturnal choking,
heavy snoring, sweating, restless sleep, impotence, and
witnessed apnea. While (OSA) is not a rare condition, it is most
likely for its victims to be middle aged or elderly. Those
affected by OSA also tend to be obese [3]. Despite how
common it is, most cases go undiagnosed and can be attributed
to 70 billion dollars loss, 11.1 billion in damages and 980
deaths each year [5].
Most cases go undiagnosed because of the inconvenience,
expenses and unavailability of testing. Testing is inconvenient
to the patient because it requires them to spend the night away
from their bed causing discomfort. It is expensive because
testing is done in the hospital, causing machines and various
technicians and staff to work over night. Testing is also widely
unavailable due to sleep centers operating at full capacity and
those on the waiting list can be untreated for an additional 6
months.
The traditional testing process includes a polysomnography
(PSG), which is a standard procedure for all sleep disorder
diagnosis. It records the breath airflow, respiratory movement,
oxygen saturation, body position, electroencephalogram
(EEG), electrooculogram (EOG), electromyogram (EMG), and
electrocardiogram (ECG) [6].
To summarize, the mere dependency on PSG needs to be
taken away from the laboratory for simpler detection and faster
treatment. Instead, automated, at-home devices that patients
can simply use while asleep seem to be very attractive and
highly on-demand.
978-1-4673-0818-2/12/$31.00 ©2012 Crown
B. Contribution
Though PSG has been widely used to detect sleep apnea,
many other techniques that rely on one biometric (e.g. ECG)
have been thoroughly investigated. To identify sleep apnea,
we propose a novel methodology in this paper that combines
RR-interval based features of the ECG signal based on the two
approaches suggested by Chazal et al., and Yilmaz et al.
Performance assessment of the combination of these two
approaches is done by measuring the classification
performance in determining the presence of apnea for different
epoch lengths.
C. Paper Organization
The rest of this paper is organized as follows. In Section II,
we glance at a variety of sleep apnea detection methods.
Section III contains an overview of the system, including a
description of the database of subjects, and details on the
analysis methodology of the paper. We describe the steps to
determine RR-interval and features extraction for different
epoch lengths in the same Section. A detailed clarification on
the SVM classifier used in our system is also provided. In
Section IV, we detail the results of our system, and then we
present a comparison with other SA detection works. Finally,
Section V concludes this paper regarding the potential
usefulness of our system, and highlights some directions for
future research.
II. R
ELATED WORK
Several methods have been suggested for identification of
sleep apnea over the past few years. Statistical features of
different signals such as nasal air flow, the thorax and abdomen
effort signals, acoustic speech signal, oxygen saturation,
electrical activity of the brain (EEG), and electrical activity of
the heart (ECG) are commonly used in the detection.
Ng et al. [7] showed that thoracic and the abdominal signals
were good parameters for the identification of the occurrence
of sleep apnea. Using the mean of absolute amplitudes of the
thoracic and the abdominal signals, they have achieved a good
performance with a receiver operating characteristic value
higher than 80%.
Depending on the hypothesis that speech signal properties
of OSA patients will be different than those not having OSA,
Goldshtein et al. [8] developed a gaussian mixture model-based
system to classify between the OSA and non-OSA groups.
They achieved a specificity and sensitivity of 83% and 79% for
the male OSA and 86% and 84% for the female OSA patients,
respectively. Their technique relied on vocal tract length and
linear prediction coefficients features.
The study in [9] assessed analysis of a comprehensive
feature set based on blood oxygen saturation (SaO
2
) from
nocturnal oximetry in order to evaluate sleep quality. The three
features of SaO
2
signal which are delta index, central tendency
measure and oxygen desaturation index are evaluated. Central
tendency measure accuracy was higher than those provided by
delta index and oxygen desaturation index. With central
tendency measure the sensitivity was 90.1% and the specificity
was 82.9%.
The relationship between periodic changes in the oxygen
saturation (SaO
2) profile and in the EEG pattern due to apnea
events during the night was investigated in [10]. The spectral
analysis of these two signals achieved 91% sensitivity, 83.3%
specificity and 88.5% accuracy in OSA diagnosis.
In [11], the authors analyze various feature sets and a
combination of classifiers based on the arterial oxygen
saturation signal measured by pulse oximetry (SpO
2
) and the
ECG in order to evaluate sleep quality and detect apnea. With
selected features of the SpO
2 and ECG signals, the Bagging
with REP Tree classifier achieved sensitivity of 79.75%,
specificity of 85.89% and overall accuracy of 84.40%.
Wavelet transforms and an artificial neural network (ANN)
algorithm were applied to the EEG signal in [12] to find a
solution to the problem of identifying sleep apnea (SA)
episodes. The system's identification results achieved a
sensitivity of approximately 69.64% and a specificity of
approximately 44.44%.
Many studies show that detection of obstructive sleep
apnea can be performed through heart rate variability (HRV)
and the ECG signal.
Quiceno-Manrique et al. [13] proposed a simple diagnostic
tool for OSA with a high accuracy (up to 92.67%) using time-
frequency distributions and dynamic features in ECG signal.
Moreover, based on spectral components of heart rate
variability, frequency analysis was performed in [14] using
Fourier and Wavelet Transformation with appropriate
application of the Hilbert Transform, where the sensitivity was
90.8%.
In addition, in [15] a bivariate autoregressive model was
used to evaluate beat-by-beat power spectral density of HRV
and R peak area, where the classification results showed
accuracy higher than 85%.
The technique in this work also relies on features of the
ECG signal.
III. METHODOLOGY
In this work, we focus on the ECG signal features to detect
sleep apnea. The block diagram of the overall methodology
used in this study is shown in Figure 1.
A. Subjects
The database of ECG signals used is available from the
PhysioNet web site [16]. PhysioNet contains a growing
collection of biomedical signals from healthy subjects and
patients. The PhysioNet web site is a public service of the
PhysioNet Resource funded by the National Institutes of
Health's NIBIB and NIGMS.
PhysioNet offers free access to Apnea-ECG Database,
which we used to assess and validate our approach. The Apnea-
ECG Database contains 70 recordings, containing a single
ECG signal varying in length from slightly less than 7 hours to
nearly 10 hours each. The sampling frequency used for ECG
acquisition was 100 Hz, with 16-bit resolution, and one sample
bit representing 5µV. The standard sleep laboratory ECG
electrode positions were used (modified lead V2) [6].
Figure 1. Schematic diagram of the system.
The subjects of these recordings were men and women
between 27 and 63 years of age (mean: 43.8±10.8 years) with
weights between 53 and 135 kg (mean: 86.3±22.2 kg). The
sleep recordings originated from 32 subjects (25 men, 7
female) those who were recruited for previous studies on
healthy volunteers, and also patients with obstructive sleep
apnea [6].
B. ECG
The electrocardiogram is a representation of the electrical
activity of the heart; each activity has a distinctive waveform.
Normal ECG graph consists of the P wave, QRS complex and
the T wave. A small U wave is normally visible in 50-75% in
the ECG [17]. Figure 2 shows a schematic representation of
normal ECG.
ECG is considered as one of the most efficient features to
detect sleep disorders. Cyclic variations in the duration of a
heartbeat, also known as RR intervals (time interval from one
R wave to next R wave) of ECG have been reported to be
associated with sleep apnea episodes. This consists of
bradycardia during apnea followed by tachycardia upon its
cessation [6].
According to [17], RR interval time series is generated for
each ECG beat, as follows:
)1(.1,...,2,1),()1()( =+= niiririrr
RR-interval is defined as the time interval between two
consecutive R peaks.
Figure 2. Schematic representation of normal ECG.
Several researches have been conducted to recognize sleep
apnea using the features derived from the RR interval such as
median, mean, inter-quartile range (IQR), and the standard
deviation of the change in RR intervals [6][17][18].
C. Data Preparation
To select the data, we chose the ECG records which have
continuous apnea data for a certain period of time, followed by
a regular (normal) data representation for a period of time, or
vice versa. The data preparation is used for training the SVM
classifier (see subsection III.G).
To give more clarification about data preparation in our
work, an example is provided regarding the data selection. In
data set1 of record a03, to get regular data, we chose the data
from 2:27:00.000 to 2:57:00.000, and to get apnea data, we
chose the data from 3:06:00.000 to 3:36:00.000. The reason of
choosing those periods was because the data at those periods
have clear apnea and regular data.
In data set2 of record b02, to get apnea data, we chose
apnea data from 1:17:00.000 to 1:37:00.000, and to get regular
data, we chose regular data from 1:57:00.000 to 2:17:00.000.
Similarly, the reason of choosing those periods was because
the data at those periods had clear apnea and regular data, and
they are within the same hour.
MATLAB toolset was used in our experiments for signal
processing. The data records were imported as MATLAB
matrices (.mat) from the Physionet web site.
The next step in our procedure after data selection is data
partitioning. In our work, three cases of partitioning were
analyzed, as follows:
Case 1. The apnea and regular data are partitioned
into 10 second pieces
Case 2. The apnea and regular data are partitioned
into 15 second pieces
Case 3. The apnea and regular data are partitioned
into epochs of 30 second pieces.
Since apnea is defined as a pause in breathing, and can last
from a few seconds to minutes (almost >=10 sec); we
investigate the three above cases to determine the best accuracy
that can be achieved.
ECG signal
Data Preparation
RR Interval Detection
Features Extraction
Data Randomization
SVM
Performance Evaluation
Figure 3. Detection of R- Peak.
D. RR Interval Detection
We need to distinguish the R waves from the other waves
of the ECG signal. Therefore, we developed the following two
conditions, in which R-peak was detected. An R peak will be
identified if both conditions 1 and 2 are satisfied:
1) It has to be a local maximum, which is detected by a
local max function within a window of 150ms.
2) The local max peaks must be at least 2 standard
deviation above the mean.
Once the R-peak was determined, RR intervals were
computed. The RR interval is the peak to peak time period
from two continuous peak signals as shown in Equation 1.
Figure 3 shows the detection of R-peaks.
E. Features Extraction
Our technique relies on an effective combination of ECG
signal features which is a novel hybrid of features extracted
from [6] and [19]. According to [17], the following ECG
features which are most effective for apnea detection are
calculated:
Mean epoch and recording RR-interval.
Standard deviation of the epoch and recording RR-
interval.
The NN50 measure (variant 1), defined as the number
of pairs of adjacent RR- intervals where the first RR-
interval exceeds the second RR- interval by more
than 50 ms.
The NN50 measure (variant 2), defined as the number
of pairs of adjacent RR-intervals where the second
RR-interval exceeds the first RR interval by more
than 50 ms.
Two pNN50 measures, defined as each NN50
measure divided by the total number of RR-intervals.
The SDSD measures, defined as the standard
deviation of the differences between adjacent RR-
intervals.
The RMSSD measures, defined as the square root of
the mean of the sum of the squares of differences
between adjacent RR- intervals.
Median of RR-intervals.
Inter-quartile range, defined as difference between
75
th
and 25
th
percentiles of the RR-interval value
distribution.
Mean absolute deviation values, defined as mean of
absolute values obtained by the subtraction of the
mean RR-interval values from all the RR-interval
values in an epoch.
The first seven features are proposed by Chazal et al. [6],
while the three latter features are proposed by Yilmaz et al.
[19], who claimed that RR interval mean, standard deviation,
and range are sensitive to outliers, and thus classification
performance deteriorates when only these features are
included.
Our hybrid technique includes a combination of the most
effective set of RR-interval based features of the ECG signal
for classification. The classification results confirm the
improved accuracy compared to the two above techniques.
F. Data Randomization
In this step, we separate the apnea and non apnea data. We
then separate training data and testing data, with 80% for the
training and 20% for the testing.
We use a MATLAB built-in function (rand) to determine
whether a feature set in 10s (or 15s or 30s) of data belongs to
test group or rule creation group. If ‘rand’ is larger than 0.2,
then the 10s data will belong to rule creation group, otherwise
it will belong to testing group.
After the signals are separated, we perform the training
for SVM.
G. Support Vector Machines
We use Support Vector Machines (SVMs) as a
classification (also known as supervised learning) method in
order to investigate apneaic epoch detection.
SVMs are learning methods, which aim to find the optimal
separating plane that analyze data and recognize pattern used
for regression analysis.
In SVM, P data is classified to which class it belongs, by
points with a (P 1) dimensional hyperplane, which is called
a linear classifier. The optimal hyperplane that separates the
clusters of vectors is found by SVM modeling. The cases with
one category of the target variable are on one side of the plane
and cases with the other category are on the other side of the
plane. Figure 4 illustrates the working principle of SVM.
A good separation between the two possible classes is
achieved by building a maximal margin hyperplane. The
margin maximizes the distance between the classes and the
nearest data point of each class. In general, the larger the
margin is, the lower the generalization error of the classifier
[17]. Figure 5 shows the trade off margin choice.
In addition, SVMs handles the separation by a kernel
function to map the data into a different space with a
hyperplane. SVM gives the flexibility for the choice of the
kernel, as shown in Figure 6. Linear, polynomial and radial
can be taken as an example for a kernel function.
Figure 4. The SVM algorithm.
Figure 5. Trade off margin choice.
Figure 6. Kernal choice.
The choice of a kernel depends on the problem we are
trying to model. Polynomial kernels are well suited for
problems where all the training data is normalized, and it
allows to model feature conjunctions up to the order of the
polynomial. Radial basis functions allow picking out circles
(or hyperplanes). In contrast, the linear kernel, allows only
picking out lines (or hyperplanes) [20].
In our implementation, we use a linear kernel function to
map the training data into kernel space. In the optimization
process, we use a method called sequential minimal
optimization to find the separating hyperplane.
IV.
RESULTS
A. Performance Evaluation
We evaluated the effectiveness of our model on the Apnea-
ECG database, using different records available in that
database. The model was implemented using MATLAB
toolset.
To evaluate the performance of the classification system,
two statistical indicators, Sensitivity (Se) and Specificity (Sp)
in addition to the Accuracy (Acc) have been used. The
sensitivity of a test is the percentage of patients in the OSA
positive group correctly diagnosed, whereas the specificity is
the percentage of subjects in the OSA negative group correctly
classified by the test.
Table I, II and III show the classification results for the
three cases mentioned in the data partitioning step. Our model
was based on a linear kernel SVM using various RR-interval
features of the ECG signal. The three cases used here are:
(i)
10 seconds data partitioning, (ii) 15 seconds, and (iii) 30
seconds.
The accuracy of our approach is 86.1%, 96.5%, and 95%,
respectively.
From Table II, SVM with linear kernel using 15 second
epochs shows the best classification accuracy with high
successful rate of correct prediction.
TABLE I
10 sec. (Accuracy is 86.1%)
Input\Output Regular Apnea
Regular
97.2%
2.78%
Apnea 25%
75%
TABLE II
15 sec. (Accuracy is 96.5%)
Input\Output Regular Apnea
Regular
100%
0%
Apnea 7.1%
92.9%
TABLE III
30 sec. (Accuracy is 95%)
Input\Output Regular Apnea
Regular
100%
0%
Apnea 10%
90%
B. Comparison with other techniques
We performed a comparison with other SA detection
works. Table IV represents comparative results. As can be
seen, our system has achieved a comparable or better
performance. This applies to the other works that rely on the
ECG signal as well as other biometric signals.
Small Margin
Large Margin
Small Margin Large Margin
Support Vectors
Input Space Input Space
Features Space
Marg
i
n
Ma
pp
in
g
Solution
TABLE IV
Comparison of Sleep Apnea Detection Approaches
Method Ref. Approach
Perform
a
nce
[%]
S
S
p
Acc.
Chazal et al. [6] Measure of minutes of
sleep disordered
respiration
NA NA 91
Alvarez et al. [9] SaO2 90.1 82.9 NA
Alvarez et al. [10] SaO2 and EEG signal 91 83.3 88.5
Xie et al. [11] SpO2 and ECG 79.75 85.89 84.40
Lin et al. [12] EEG signal 69.64 44.44 NA
Quiceno-
Manrique et
al.
[13] ECG signal NA NA 92.67
Schrader et
al.
[14] Fourier and Wavelet
Transformation of HRV
90.8 NA NA
Mendez et al. [15] Bivariate autoregressive
model of HRV
NA NA 85
Yilmaz et al. [19] RR-interval based
classification
NA NA 89
Proposed - Features extraction of
ECG signal
92.9 100 96.5
V. CONCLUSIONS AND FUTURE WORKS
In this work, we studied the possibility of the detection of
sleep apnea or hypopnea events from the ECG signal variation
patterns during sleep. We further developed a model using the
ECG signal features and evaluated its effectiveness. Our model
was based on a selective set of RR-interval-based features that
were given to an SVM for classification. We evaluated our
model on three different epoch lengths.
From the experimental results, we conclude that SVM with
linear kernel shows the best accuracy with 15 second epoch
length.
As a future work, we plan to do performance optimization
for feature selection, and then incorporate this work into a real-
time monitoring system that acquires and analyzes the ECG
signal of subjects during sleep.
R
EFERENCES
[1] Sleep Disorder Overview. www.neurologychannel.com.
[2] S. Isa, M. Fanany, W. Jatmiko and A. Arymurthy, “Sleep Apnea
Detection from ECG Signal, Analysis on Optimal Features, Principal
Components, and Nonlinearity,” in Proceedings of the 5
th
IEEE
International Conference on Bioinformatics and Biomedical
Engineering (iCBBE), pp. 1-4, May 2011.
[3] Sleep Apnea: What Is Sleep Apnea? www.nhlbi.nih.gov.
[4] Apnea guide. www.apneaguide.com.
[5] Treating Sleep Apnea Could Cut Road Deaths.
www.americanvoiceinstitute.org.
[6] P. Chazal, T. Penzel and C. Heneghan,“Automated Detection of
Obstructive Sleep Apnoeaa at Different Time Scales Using the
Electrocardiogram,” Institute of Physics Publishing, vol. 25, no. 4, pp.
967-983, Aug. 2004.
[7] A. Ng, J. Chung, M. Gohel, W. Yu, K. Fan and T. Wong, “Evaluation of
the Performance of Using Mean Absolute Amplitude Analysis of
Thoracic and Abdominal Signals for Immediate Indication of Sleep
Apnoea Events,” Journal of Clinical Nursing, vol. 17, no. 17, pp. 2360-
2366, Sep. 2008.
[8] E. Goldshtein, A. Tarasiuk and Y. Zigel, “Automatic Detection of
Obstructive Sleep Apnea Using Speech Signals,” in IEEE Transactions
on Biomedical Engineering, vol. 58, no. 5, pp. 1373-1382, May. 2011.
[9] D. Alvarez, R. Hornero, D. Abasolo, F. Campo and C. Zamarron,
“Nonlinear Characteristics of Blood Oxygen Saturation from Nocturnal
Oximetry for Obstructive Sleep Apnoea Detection,Institute of Physics
Publishing, vol. 27, no. 4, pp. 399-412, Apr. 2006.
[10] D. Avarez, R. Hornero, J. Marcos, F. Campo and M. Lopez, “Spectral
Analysis of Electroencephalogram and Oximetric Signals in Obstructive
Sleep Apnea Diagnosis,” in Proceedings of the 31
st
IEEE International
Conference on Engineering in Medicine and Biology Society (EMBS
2009), pp. 400-403, Sep. 2009.
[11] B. Xie, H. Minn, “Real Time Sleep Apnea Detection by Classifier
Combination,” in IEEE Transactions on Information Technology in
Biomedicine (in Press), 2012.
[12] R. Lin, R. Lee, C. Tseng, H. Zhou, C. Chao, J. Jiang,A New Approach
for Identifying Sleep Apnea Syndrome Using Wavelet Transform and
Neural Networks,” Biomedical Engineering: Applications, Basis &
Communications, vol. 18, no. 3, pp. 138-143, 2006.
[13] Q. Manrique, A. Hernandez, T. Gonzalez, F. Pallester and C.
Dominquez, “ Detection of Obstructive Sleep Apnea in ECG Recordings
Using Time-Frequency Distributions and Dynamic Features,” in
Proceedings of the IEEE International Conference on Engineering in
Medicine and Biology Society( EMBS 2009), pp. 5559-5562, Sep. 2009.
[14] M. Schrader, C. Zywietz, V. Einem, B. Widiger, G. Joseph, “Detection
of Sleep Apnea in Single Channel ECGs from the PhysioNet Data
Base,” Computers in Cardiology 2000, vol. 27, pp. 263-266, Sep. 2000.
[15] M. Mendez, D. Ruini, O. Villantieri, M. Matteucci, T. Penzel and A.
Bianchi, “Detection of Sleep Apnea from Surface ECG Based on
Features Extracted by an Autoregressive Model, ” in Proceedings of the
IEEE International Conference on Engineering in Medicine and Biology
Society (EMBS 2007), pp. 6105-6108, Aug. 2007.
[16] PhysioNet, www.physionet.org.
[17] S. Isa, M. Fanany, W. Jatmiko and A. Murini, “Feature and Model
Selection on Automatic Sleep Apnea Detection Using ECG,” in
International Conference on ComputerScience and Information Systems,
ICACSIS 2010, pp. 357-362
, 2010.
[18] P. Langley, E. Bowers and A. Murray,Principal Component Analysis
as Tool for Analyzing Beat-to-Beat Changes in ECG Features:
Application To ECG-Derived Respiration,” in IEEE Transactions on
Biomedical Engineering, vol. 57, no. 4, pp. 821-829, Apr. 2010.
[19] B. Yilmaz, M. Asyali, E. Arikan, S. Yektin and F. Ozgen, “Sleep Stage
and Obstructive Apneaic Epoch Classification Using Single-Lead ECG,”
in Biomedical Engineering Online, vol. 9, 2010.
[20] A. Jain, P. Flynn and A. Ross, “Handbook of Biometrics,” Springer,
New York, 2008.
... The most common method for apnea diagnosis is ECG [5]. ECG records the electronic signals generated from the human heart. ...
... Although there are many statistical body measures like ECG, acoustic speech signal, Sa, Electroencephalogram (EEG) available for apnea diagnosis [5], we have solely focused on ECG signal for our work. A lot of research works on apnea diagnosis from ECG signals have already been performed. ...
... A lot of research works on apnea diagnosis from ECG signals have already been performed. Almazaydeh et al. [5] have extracted the relevant statistical features such as mean, standard deviation, median, inter-quartile range and some of their derivations for an RR interval (interval between two consecutive R peaks) of the raw ECG signals of the PhysioNet Apnea-ECG database [12]. They have applied SVM on these extracted features and have achieved a maximum of 96.5% accuracy. ...
Article
Full-text available
Sleep Apnea is a breathing disorder occurring during sleep. Older people suffer most from this disease. In-time diagnosis of apnea is needed which can be observed by the application of a proper health monitoring system. In this work, we focus on Obstructive Sleep Apnea (OSA) detection from the Electrocardiogram (ECG) signals obtained through the body sensors. Our work mainly consists of an experimental study of different ensemble techniques applied on three deep learning models-two Convolutional Neural Network (CNN) based models, and a combination of CNN and Long Short-Term Memory (LSTM) models, which were previously proposed in the OSA detection domain. We have chosen four ensemble techniques-majority voting, sum rule and Choquet integral based fuzzy fusion and trainable ensemble using Multi-Layer Perceptron (MLP) for our case study. All the experiments are conducted on the benchmark PhysioNet Apnea-ECG Database. Finally, we have achieved highest OSA detection accuracy of 85.58% using the MLP based ensemble approach. Our best result is also able to surpass many of state-of-the-art methods.
... kg. The sleep recordings were obtained from 25 male and 7 female volunteers, including both healthy and OSA subjects [41,42]. ...
... STFT is used to construct the TFR of the physiological signals as a spectrogram with a constant time-frequency resolution (see Section 2, Eq 1). A constant sliding window along the time axis is employed to create a two-dimensional (2D) representation of the signal at this fixed resolution [41,44,45]. As a result of using a constant window, all the frequency information is analyzed at the same time-frequency resolution. ...
Article
Full-text available
Obstructive sleep apnea (OSA) is a common chronic sleep disorder that disrupts breathing during sleep and is associated with many other medical conditions, including hypertension, coronary heart disease, and depression. Clinically, the standard for diagnosing OSA involves nocturnal polysomnography (PSG). However, this requires expert human intervention and considerable time, which limits the availability of OSA diagnosis in public health sectors. Therefore, electrocardiogram (ECG)-based methods for OSA detection have been proposed to automate the polysomnography procedure and reduce its discomfort. So far, most of the proposed approaches rely on feature engineering, which calls for advanced expert knowledge and experience. This paper proposes a novel fused-image-based technique that detects OSA using only a single-lead ECG signal. In the proposed approach, a convolutional neural network extracts features automatically from images created with one-minute ECG segments. The proposed network comprises 37 layers, including four residual blocks, a dense layer, a dropout layer, and a soft-max layer. In this study, three time-frequency representations, namely the scalogram, the spectrogram, and the Wigner-Ville distribution, were used to investigate the effectiveness of the fused-image-based approach. We found that blending scalogram and spectrogram images further improved the system's discriminative characteristics. Seventy ECG recordings from the PhysioNet Apnea-ECG database were used to train and evaluate the proposed model using 10-fold cross validation. The results of this study demonstrated that the proposed classifier can perform OSA detection with an average accuracy, recall, and specificity of 92.4%, 92.3%, and 92.6%, respectively, for the fused spectral images.
... Many literature proposed techniques of sleep apnoea detection based on either ECG and SpO 2 signals alone [11,13,16,17,25,26], and in a few cases also based on a combination of ECG and SpO 2 signals [20,27]. However, in most literature, either the data used for training and testing is small or the accuracy claimed is not satisfactory which compromises the reliability of the diagnosis. ...
Article
Full-text available
Sleep apnoea is a potentially serious sleep disorder that is characterised by repetitive episodes of breathing interruptions. Traditionally, sleep apnoea is commonly diagnosed in an attended sleep laboratory setting using polysomnography (PSG). The manual diagnosis of sleep apnoea using PSG is, however complex, and time-consuming, as many physiological variables are usually measured overnight using numerous sensors attached to patients. In PSG sleep laboratories, an expert human observer is required to work overnight, and the diagnosis accuracy is dependent on the physician's experience. A quantitative and objective method is required to improve the diagnosis efficacy, decrease the complexity and diagnosis time and to ensure a more accurate diagnosis. The purpose of this study was then to develop an automatic sleep apnoea and severity classification using a simultaneously recorded electrocardiograph (ECG) and saturation of oxygen (SpO2) signals based on a machine learning algorithm. Different ECG and SpO2 time domain and frequency domain features were extracted for training different machine learning algorithms. For sleep apnoea classification, an accuracy of 99.1%, specificity of 98.1% and sensitivity of 100% were achieved using a support vector machine (SVM) based on combined ECG and SpO2 features. Similarly, for severity classification, an 88.9% accuracy, 90.9% specificity and 85.7% sensitivity have been obtained. For both apnoea and severity classification, using the combined features was found to be more accurate, and this is typically important when either channel is poor quality, the system can make an analysis based on the other channel and achieve good accuracy.
... Many machine learning-based methods were discussed for sleep apnea detection by combining various features like ECGderived respiration (EDR) and heart rate variability (HRV) derived from ECG signals [22], [28]- [30]. Convolutional neural networks have been used to generate features that can be used for sleep apnea detection in [26]. ...
Article
Full-text available
With advances in circuit design and sensing technology, the acquisition of data from a large number of sensors simultaneously to enable more accurate inferences has become mainstream. In this work, we propose a novel convolutional neural network (CNN) model for the fusion of multimodal and multiresolution 1-dimensional signals obtained from different sensors. The proposed model enables the fusion of multiresolution sensor signals, without having to resort to padding/ resampling to correct for frequency resolution differences even when carrying out temporal inferences like high-resolution event detection. The performance of the proposed model is evaluated for sleep apnea event detection, by fusing three different sensor signals obtained from UCD St. Vincent University Hospitals sleep apnea database. The generalizability of the model is demonstrated by incremental performance improvements, proportional to the number of sensors used for fusion. A selective dropout technique is used to prevent overfitting of the model to any specific high-resolution input, and increase the robustness of fusion to signal corruption from any sensor source. A fusion model with electrocardiogram (ECG), Peripheral oxygen saturation signal (SpO2), and abdominal movement signal achieved an accuracy of 99.72% and a sensitivity of 98.98%. Energy per classification of the proposed fusion model was estimated to be approximately 5.61 uJ for on-chip implementation. The feasibility of pruning to reduce the complexity of the fusion models was also studied.
... Bu çalışma kardiovasküler parametrelerin daha basit ve ucuz yöntemler ile ölçümüne dayalı olarak apne tespiti yapmaya odaklanan çalışmalar ortaya çıkmasına yol açmıştır. Tek kanallı EKG den elde edilen kalp hızı değişkenliği parametrelerini ve çeşitli sınıflandırıcılar kullanan çeşitli çalışmalarda, apne tespit başarımı % 95'lerin üzerine çıkmıştır [12][13][14]. Bazı çalışmalarda sinyal işleme teknikleri kullanılarak EKG'den elde edilen solunum sinyalleri de, apne tespitine yardımcı olmak için kullanılmıştır [15]. ...
Article
Full-text available
Nowadays, increasing obesity and overweight due to eating and drinking habits cause many people to have sleep apnea syndrome, which manifests itself in the form of snoring, by obstructing the airways during sleep. Although there are diagnostic tools such as clinical and radiological examination, the gold standard measurement method for diagnosing this disease is PolySomnoGraphy (PSG). For the diagnosis of sleep apnea with polysomnography, while the patient is sleeping in the sleep laboratory in the hospital, many physiological parameters such as ECG, EEG, EMG, EOG, air flow in the mouth or nose, the chest and diaphragm circumference should be measured and evaluated by a specialist. Due to disadvantages such as patients who have difficulty sleeping in the hospital environment, small number of laboratories performing the test and the cost of the test, a new quest has begun. In this study, new methods developed as an alternative to polysomnography were reviewed and compared. As an alternative to polysomnography, Firstly, the methods based on measuring only a few physiological parameters in one's own home have been proposed. Many studies have been conducted to detect apnea by recording one or more of the respiratory signals, ECG, photoplethysmography, or pulse oximetry signals. These methods did not reach the expected clinical usage level due to the difficulties in the patient's self-enrollment. To overcome this problem, methods based on thermal camera or ultrasonic tracking, of the person that do not require electrical contact to the person have been proposed. As an alternative to these methods that require follow-up of the patient throughout the night, studies on apnea detection from speech voice recordings taken in the hospital environment for a few minutes while the person is awake have emerged. Similar study protocols were conducted in most of the articles reviewed. The methods generally include determination of features from the recorded signals, applying them to classifiers, deciding as to whether the person has apnea or the level of apnea. Among the alternative methods to polysomnography; it has been observed that apnea detection can be achieved with a success rate of over 90% with the methods that use physiological signal recordings during the person's sleep at home. On the other hand, a success of around 80% has been achieved in apnea detection studies from voice recordings made when the person was awake. As a result, it was evaluated that the detection of apnea from voice recordings of a few minutes while the person was awake would provide great convenience for the field, but its performance should be increased, and it should be confirmed by extensive clinical studies.
... In [10], the authorsintroduced a light and small wireless sleep activity monitoringsystem, the postures and change in posture were detectedusing tri-axis accelerometer. In [11,12] OSA is diagnosedusing electrocardiogram (ECG) signals and the system can beused as a basis for future development of a tool for OSA screening. As previously discussed, the sleeping positionsare used in a number of medical applications, one of thembeing pressure ulcer. ...
... Apnea is also linked to narcolepsy, insomnia, and obesity [9]. Studies show that patients with apnea have a higher chance of being involved in a road traffic accident [10]. The disease is also a risk factor for complications during operations under anesthesia [11]. ...
Article
Sleep apnea is a common condition that is characterized by sleep-disordered breathing. Worldwide the number of apnea cases has increased and there has been a growing number of patients suffering from apnea complications. Unfortunately, many cases remain undetected, because expensive and inconvenient examination methods are formidable barriers with regard to the diagnostics. Furthermore, treatment monitoring depends on the same methods which also underpin the initial diagnosis; hence issues related to the examination methods cause difficulties with managing sleep apnea as well. Computer-Aided Diagnosis (CAD) systems could be a tool to increase the efficiency and efficacy of diagnosis. To investigate this hypothesis, we designed a deep learning model that classifies beat-to-beat interval traces, medically known as RR intervals, into apnea versus non-apnea. The RR intervals were extracted from Electrocardiogram (ECG) signals contained in the Apnea-ECG benchmark Database. Before feeding the RR intervals to the classification algorithm, the signal was band-pass filtered with an Ornstein–Uhlenbeck third-order Gaussian process. 10-fold cross-validation indicated that the Long Short-Term Memory (LSTM) network has 99.80% accuracy, 99.85% sensitivity, and 99.73% specificity. With hold-out validation, the same network achieved 81.30% accuracy, 59.90% sensitivity, and 91.75% specificity. During the design, we learned that the band-pass filter improved classification accuracy by over 20%. The increased performance resulted from the fact that neural activation functions can process a DC free signal more efficiently. The result is likely transferable to the design of other RR interval based CAD systems, where the filter can help to improve classification performance.
... This lack of awareness, beyond being a health issue, is a social and economic problem. For example, in the USA, it causes each year the loss of 70 billion dollars, 11.1 billion in damages, and 980 deaths [33]. ...
Article
Deep Neural Networks (DNNs) may be very effective for the classification over highly-sized data sets, especially in the medical domain, where the recognition of the occurrence of a specific event related to a disease is of high importance. Unfortunately, DNNs suffer from the drawback that a good set of values for their configuration hyper-parameters must be found. Currently, this is done through the use of either trial-and-error methods or sampling-based ones. In this paper we propose a new approach to find the most suitable structure for a DNN used for a classification problem in terms of achievement of the highest classification accuracy. This approach is based on a distributed version of Differential Evolution (DE), a variety of an Evolutionary Algorithm. To evaluate the approach, in this paper we investigate this issue with reference to Obstructive Sleep Apnea (OSA). OSA is an important medical problem consisting of episodes taking place during night in which a subject stops breathing due to a constriction of the upper airways. This deteriorates the quality of life and may have dangerous, and even lethal, consequences on both short and long term. An accurate classification is a very crucial step for the OSA treatment, because understanding automatically that a subject is experiencing such an episode may be decisive if prompt medical action is needed. In our experiments, classification takes place on a data set in which each item contains the values of 17 Heart Rate Variability parameters, extracted from ElectroCardiography signals, and the annotation of OSA events. We have extracted this data set from the real-world Sleep Heart Health Study database. The results obtained by the distributed DE are compared against those of the Grid Search as well as against those achieved by 13 well-known classification tools. The use of a distributed DE version turns out to be very effective in automatically obtaining DNN structures with higher classification accuracy with respect to Grid Search (72.95% versus 72.61%), and allows saving a high amount of time (three hours as opposed to 65 h and 40 min). Moreover, the proposed method outperforms in terms of higher accuracy all the other classifiers investigated, as it is evidenced also by statistical analysis. Numerically, the runner-up, i.e., JRip, achieves as its best value 72.01% and 71.50% on average over 25 runs, both values being lower than 72.95% and 72.74% obtained by our dDE.
... First, a vast amount of studies concentrate on equal-length short term heart rate variability signal modeling. The heart rate variability signals are divided into equal-length short term signals, from which a variety of features including frequency domain [7,8], time domain [9] and nonlinear domain features [10] are derived. Afterwards, statistical diagnostic techniques fed by features are applied to detect the disease state of that short term signal. ...
... The brain and rest of the body may not get sufficient oxygen for sleep apnea patients. Sleep apnea is of three types: obstructive, central and mixed sleep apneas [10,11]. Obstructive sleep apnea is caused mainly due to blockage of airflow. ...
Conference Paper
Full-text available
The purpose of this study is to find optimal features and classifier's model selection for sleep apnea detection using ECG signals. We want to determine whether a set of unknown ECG signals (test data) is from heavy apnea, mild apnea, or healthy categories. We examine two recent approaches of features selection: an approach proposed by Chazal et al. (2004), which is based on the RR-interval mean and time-series analysis; and an approach proposed by Yilmaz et al. (2010), which is based on the RR-interval median. We also examine cross validation and random sampling method in the classifier's probability model selection. We evaluate the approaches using three classifiers: k-Nearest Neighbor (kNN), Naive-Bayes and Support Vector Machine (SVM). In addition, we use a self organizing map (SOM) clustering or preprocessing to provide better sample that can represent the entire training data. Our experiment using ECG data from PhysioNet shows that classification results using only 3 features as proposed by Yilmaz et al. (2010) gives about 3.59% gain on overall classification accuracy (CA) and 7.5% gain on area under ROC-curve (AUC) on than the classification accuracy using 8 features as proposed by Chazal et al., (2004).
Article
Full-text available
This paper describes implementation of Principal Component Analysis (PCA) on sleep apnea detection using Electrocardiogram (ECG) signal. The statistics of RR-intervals per epoch with 1 minute duration were used as an input. The combination of features proposed by Chazal and Yilmaz was transformed into orthogonal features using PCA. Cross validation, random sampling, and test on train data were used on model selection. The results of classification using kNN, Naïve- Bayes, and Support Vector Machine (SVM) show that PCA features give better classification accuracy compared to Chazal and Yilmaz features. SVM with RBF (Radial Basis Function) kernel gives the best classification accuracy by using 7 principal components (PC) as a features. The experimental results show that relation between Chazal features with target class tend to be linear, but Yilmaz and PCA features are non-linear.
Article
Full-text available
This paper describes a new technique to classify and analyze the electroencephalogram (EEG) signal and recognize the EEG signal characteristics of Sleep Apnea Syndrome (SAS) by using wavelet transforms and an artificial neural network (ANN). The EEG signals are separated into Delta, Theta, Alpha, and Beta spectral components by using multi-resolution wavelet transforms. These spectral components are applied to the inputs of the artificial neural network. We treated the wavelet coefficient as the kind of the training input of artificial neural network, might result in 6 groups of wavelet coefficients per second signal by way of characteristic part processing technique of the artificial neural network designed by our group, we carried out the task of training and recognition of SAS symptoms. Then the neural network was configured to give three outputs to signify the SAS situation of the patient. The recognition threshold for all test signals turned out to have a sensitivity level of approximately 69.64 and a specificity value of approximately 44.44 . In neurology clinics, this study offers a clinical reference value for identifying SAS, and could reduce diagnosis time and improve medical service efficiency.
Article
Full-text available
Obstructive sleep apnea (OSA) is a common disorder associated with anatomical abnormalities of the upper airways that affects 5% of the population. Acoustic parameters may be influenced by the vocal tract structure and soft tissue properties. We hypothesize that speech signal properties of OSA patients will be different than those of control subjects not having OSA. Using speech signal processing techniques, we explored acoustic speech features of 93 subjects who were recorded using a text-dependent speech protocol and a digital audio recorder immediately prior to polysomnography study. Following analysis of the study, subjects were divided into OSA (n=67) and non-OSA (n=26) groups. A Gaussian mixture model-based system was developed to model and classify between the groups; discriminative features such as vocal tract length and linear prediction coefficients were selected using feature selection technique. Specificity and sensitivity of 83% and 79% were achieved for the male OSA and 86% and 84% for the female OSA patients, respectively. We conclude that acoustic features from speech signals during wakefulness can detect OSA patients with good specificity and sensitivity. Such a system can be used as a basis for future development of a tool for OSA screening.
Article
Full-text available
Polysomnography (PSG) is used to define physiological sleep and different physiological sleep stages, to assess sleep quality and diagnose many types of sleep disorders such as obstructive sleep apnea. However, PSG requires not only the connection of various sensors and electrodes to the subject but also spending the night in a bed that is different from the subject's own bed. This study is designed to investigate the feasibility of automatic classification of sleep stages and obstructive apneaic epochs using only the features derived from a single-lead electrocardiography (ECG) signal. For this purpose, PSG recordings (ECG included) were obtained during the night's sleep (mean duration 7 hours) of 17 subjects (5 men) with ages between 26 and 67. Based on these recordings, sleep experts performed sleep scoring for each subject. This study consisted of the following steps: (1) Visual inspection of ECG data corresponding to each 30-second epoch, and selection of epochs with relatively clean signals, (2) beat-to-beat interval (RR interval) computation using an R-peak detection algorithm, (3) feature extraction from RR interval values, and (4) classification of sleep stages (or obstructive apneaic periods) using one-versus-rest approach. The features used in the study were the median value, the difference between the 75 and 25 percentile values, and mean absolute deviations of the RR intervals computed for each epoch. The k-nearest-neighbor (kNN), quadratic discriminant analysis (QDA), and support vector machines (SVM) methods were used as the classification tools. In the testing procedure 10-fold cross-validation was employed. QDA and SVM performed similarly well and significantly better than kNN for both sleep stage and apneaic epoch classification studies. The classification accuracy rates were between 80 and 90% for the stages other than non-rapid-eye-movement stage 2. The accuracies were 60 or 70% for that specific stage. In five obstructive sleep apnea (OSA) patients, the accurate apneaic epoch detection rates were over 89% for QDA and SVM. This study, in general, showed that RR-interval based classification, which requires only single-lead ECG, is feasible for sleep stage and apneaic epoch determination and can pave the road for a simple automatic classification system suitable for home-use.
Conference Paper
Full-text available
Detection of obstructive sleep apnea can be performed through heart rate variability analysis, since fluctuations of oxygen saturation in blood cause variations in the heart rate. Such variations in heart rate can be assessed by means of time-frequency analysis implemented with time-frequency distributions belonging to Cohen's class. In this work, dynamic features are extracted from time frequency distributions in order to detect obstructive sleep apnea from ECG signals recorded during sleep. Furthermore, it is applied a methodology to measure the relevance of each dynamic feature, before the implementation of k-nn classifier used to recognize the normal and pathologic signals. As a result, the proposed method can be applied as a simple diagnostic tool for OSA with a high accuracy (up to 92.67%) in one-minute intervals.
Article
To find an efficient and valid alternative of polysomnography (PSG), this paper investigates real-time sleep apnea and hypopnea syndrome (SAHS) detection based on electrocardiograph (ECG) and saturation of peripheral oxygen (SpO(2)) signals, individually and in combination. We include ten machine-learning algorithms in our classification experiment. It is shown that our proposed SpO (2) features outperform the ECG features in terms of diagnostic ability. More importantly, we propose classifier combination to further enhance the classification performance by harnessing the complementary information provided by individual classifiers. With our selected SpO(2) and ECG features, the classifier combination using AdaBoost with Decision Stump, Bagging with REPTree, and either kNN or Decision Table achieves sensitivity, specificity, and accuracy all around 82% for a minute-based real-time SAHS detection over 25 sleep-disordered-breathing suspects' full overnight recordings.
Article
This study assessed the hypothesis that blood oxygen saturation (SaO(2)) and electroencephalogram (EEG) recordings could provide complementary information in the diagnosis of the obstructive sleep apnea (OSA) syndrome. We studied 148 patients suspected of suffering from OSA. Classical spectral parameters based on the relative power in specified frequency bands (A(f-band)) or peak amplitudes (PA) were used to characterize the frequency content of SaO(2) and EEG recordings. Additionally, the median frequency (MF) and the spectral entropy (SE) were applied to obtain further spectral information. We applied a forward stepwise logistic regression (LR) procedure with crossvalidation leave-one-out to obtain the optimum spectral feature set. Two features from the oximetric spectral analysis (PA and MFsat) and three features from the EEG spectral analysis (A(delta), A(alpha) and SEeeg) were automatically selected. 91.0% sensitivity, 83.3% specificity and 88.5% accuracy were obtained. These results suggest that MF and SE could provide additional information to classical frequency characteristics commonly used in OSA diagnosis. Additionally, nocturnal SaO(2) and EEG recordings during the whole night could provide complementary information to help in the detection of OSA syndrome.
Article
To evaluate the sensitivity of mean absolute amplitudes of the thoracic and the abdominal signals as a prompt indicator of the occurrence of sleep apnoea events. To provide symptomatic management of sleep apnoea, a reliable method of detecting sleep apnoea is essential to ensure that the intervention can be applied only when needed. It is also crucial to identify the threshold for the trigger of an intervention using a deployed sensor. Twenty-six subjects aged between 18-65 years who were diagnosed with obstructive or central sleep apnoea underwent an overnight sleep study. Signals of nasal and oral airflow, thoracic and abdominal efforts and pulse oximetry level were recorded using a polysomnography device. With a 95% CI, the overall area under the receiver operating characteristic of the thoracic signal, the abdominal signal and the combination of the thoracic and the abdominal signals were 84.56, 87.48 and 90.91%, respectively. Using -20, -25 and -30% as a cut-off point, the sensitivity values of thoracic signal, abdominal signal and combination of the thoracic and the abdominal signals ranged from 70.29-86.25% and the specificity values ranged from 74.82 to 90.09%. Using mean absolute amplitude analysis, the results of this study showed that combination of the thoracic and the abdominal signals achieved the best overall and individual performances compared with thoracic signal and abdominal signal. Overall, thoracic signal, abdominal signal and combination of the thoracic and the abdominal signals have a good performance with an receiver operating characteristic value higher than 80%. The thoracic and the abdominal signals were good parameters for the identification of the occurrence of sleep apnoea, being as quick as the nasal airflow signal. These results suggested that sleep apnoea events could be identified through constant monitoring of the patient's thoracic and abdominal signals. Knowledge of these signals could help nurses to manage sleep apnoea in patients.