Abstract
This paper presents the development of a sleep monitor to provide a comfortable way of detecting sleep-related breathing disorders like the obstructive sleep apnea syndrome (OSAS). OSAS is traditionally diagnosed using polysomnography, which requires a whole night stay at the sleep laboratory of a hospital with multiple electrodes and sensors attached to the patient’s body. However, body sound and motion tracking also provide extensive information about sleep course. A unique recording device offering a good body sound extraction, noise suppression and a small size is developed. Using this device a reliable detection of breathing and heart beat is possible. In addition sleeping positions and the activity of the patient will be evaluated using an inertial measurement unit (IMU). The device is easy to set up and offers the possibility to use it independently at home.
Initial experiments have shown that volunteers were able to set up the device on their own. Furthermore several overnight recordings revealed the capability to monitor breathing, heart rate, sleeping position as well as movements of the patient.
1 Introduction
The most common sleep-related breathing disorder is the obstructive sleep apnea syndrome (OSAS). OSAS is characterized by repetitive pauses in breathing. The pharynx narrows and the path of the air flow is obstructed which can lead to an obstructive apnea [1]. During an apnea the breathing amplitude decreases by 80% in respect to the baseline over a period of more than 10 seconds. Normally an initial drop in heart rate and a decrease of O2 saturation occurs a few seconds afterwards. An alarm signal to the central nervous system, the so-called arousal, terminates the apnea phase. During this period the patients heart rate increases and vigorous breathing starts. About 2-4% of the adult population suffer from OSAS with clinical symptoms. The consequences of this disease include heart disease, elevated blood pressure and extreme daytime sleepiness. Despite this high prevalence, most patients are either un-diagnosed or untreated [1].
The diagnostic standard is the polysomnography, which involves a detailed and elaborate diagnosis in a clinical sleep laboratory. Here a reliable statement is only possible with great technical effort and complexity. The sensors in sleep laboratory are connected by tubes or cables with different devices. This method is still expensive and may affect sleep which could result in a measurement bias. Therefore the early diagnosis of sleep-related respiratory disorders and the comfortable monitoring of sleep course would open up a wide market of application already starting in homecare area.
2 Development of the sleep monitor
2.1 Body sound
A considerable amount of literature has been published on sleep monitoring based on human body sound. These studies pay particular attention to recording breathing sounds near the trachea. This assumption is based on the fact that the lower neck is one of the best positions for recording breathing sounds [2]. Therefore we set up our microphone at the suprasternal notch to get the best results in recording breathing sounds. A seminal study by [3] showed that it is possible to detect sleep apnea by placing a stethoscope like microphone at the suprasternal notch over the night. A more recent study by [4] approved these results.
One of the greatest challenges is to provide a reliable and comfortable method to record body sounds. Therefore a unique recording device offering optimized body sound extraction and noise suppression, a small size and a changeable membrane for hygienic reasons is developed. Because of its compact geometry and low cost a electret microphone is used to record body sound. To fix the device at the suprasternal notch it is adhered with a plaster (see Figure 1). In order to ensure a good body sound signal, the membrane needs to maintain contact with the skin for the entire night. Therefore the used plaster must not loosen itself even if the user is moving during the night. In this work 3M Medipore fixation fleece-plasters are used. These plasters offer a high skin compability as well as a strong adhesion to the human skin.
Sensor attached to volunteers suprasternal notch.
The analog preprocessing of the microphone signal is essential for subsequent digital analysis. The frequency response of the circuit for analog preprocessing of the microphone signal is shown in Figure 2. The majority of breathing sounds feature frequencies from 100 Hz up to 1.5 kHz [2]. Therefore these frequencies get amplified by about 26 dB. Most heart- and snoring sounds feature frequencies below 100 Hz. Because those sounds are inherently very loud a smaller amplification suffices. Frequencies below 10 Hz or above 1.5 kHz feature mostly noise and sounds created by artifacts. The audio data is sampled at 5 kHz in 10 bit samples.
Frequency response of the analog preprocessing unit.
2.2 Motion tracking
An inertial measurement unit (IMU) MPU-6000 (by InvenSense) delivers position and motion data as three gyroscope values and three accelerometer values sampled at 200 Hz. As described in previous work [5] we are able to determine the sleeping position as well as the movements of the patient using the IMU data. Here stable results are provided by using the Madgwick-Filter [6]. In the majority of cases movements create audio artifacts. In general movements during sleep provide valuable information about sleep quality. Since movements are often the cause of artifacts in the audio signal, they can be used to detect and suppress those disturbances. Therefore the IMU data is essential for subsequent audio analysis.
2.3 Device setup
As shown in Figure 3, the proposed device consists of two parts. The first is carried by the patient during sleep and consists of the body sound microphone, IMU and the remaining electronics (battery, Bluetooth gateway, etc.). These are stored within a case which is carried by the patient using an abdominal belt. A single cable connects the electronics with the body sound microphone. To avoid further wiring the proposed system can be charged by inductive charging using the Qi-standard.
The second part is the masterstation, which consists of a laptop containing a user optimized software. Data is send wireless to the laptop via Bluetooth. Received data is processed, visualized and stored using the developed software. To be able to exchange and compare the recorded data, all data is stored using the European Data Format (EDF), a standard format for exchange of digitized poly-graphic recordings [7]. The subsequent automated signal analysis for the extraction of diagnostic features is currently under development.
Abstract representation of the entire sleep monitoring system. The dashed module is the subject of current developing efforts.
To set up the system the patient puts on the abdominal belt and stores the electronics within. Then the patient adheres the microphone to his suprasternal notch using the provided plaster. Afterwards the microphone cable gets connected to the electronics and the recording can be started via the software.
3 Results
The proposed system was tested on seven volunteers (age: 24-61 years, 2 female, 5 male). Each volunteer carried out a whole overnight recording at home. Participants were given instructions regarding the system setup and were asked to take the device home to use it independently.
3.1 Ergonomics and ease of use
In addition to the overnight recording each volunteer answered four questions (see Table 1) about his/her experience using the device. Answers were given qualitatively as one of four options: poor, fair, good, or excellent, each of which was quantified to a numerical value 1–4 respectively.
Ease-of-use and comfort questionnaire
| Aspect | Numerical value ± (SD) | Equivalent rating |
|---|---|---|
| Device handling | 3.0 ± 0.00 | good |
| Software ease of use | 4.0 ± 0.00 | excellent |
| Overall comfort | 3.3 ± 0.49 | good |
| Sleep quality | 3.4 ± 0.53 | good |
3.2 Body sound
All volunteers were able to setup the device and to deliver a complete and evaluable overnight recording. Figure 4 shows time and frequency representation of a recorded audio sample. Here, breathing and heart sounds are within the expected frequency ranges. This observation is very similar in all seven overnight recordings. Therefore it is a simple task to extract the breathing as well as heart sounds out of the raw signal.
An example of extracted breath sounds can be seen in the top graph of Figure 5. It shows the same audio signal as shown in Figure 4 after filtering with a band-pass filter ranging from 100 Hz up to 1.5 kHz. The previously nearly unrecognizable breathing sounds in time domain are now clearly visible. Inspiration, expiration and a silence interval in between can be recognized.
Tracheal sound signal acquired with the proposed system. The graphs show a 30 second segment of the recorded audio data during a overnight recording in time (top) and frequency (bottom) domain.
In correspondence, the bottom graph of Figure 5 shows the heart signal extracted by applying a band-pass filter ranging from 15 Hz up to 80 Hz. Additionally the characteristic cardiac sound pattern consisting of two isolated peaks per heart beat can be seen. By detecting these specific peaks the heart rate can be calculated, which provides valuable diagnostic information.
Tracheal breath sound (top) and heart sound (bottom) acquired with the proposed system after respective bandpass filtering.
The overnight recordings included two volunteers suffering from sleep apnea. Figure 6 shows the time and frequency representation of a 50 second segment of the recorded audio signal during a apnea phase. A typical apnea cycle can be observed in both time and frequency domain. The breathing sounds decrease slowly in amplitude until the breathing is nearly ceased. After about 10 seconds the apnea phase is terminated with heavy and short-winded breathing.
This example demonstrates, that it is already possible to extract diagnostic information by visual examination of the filtered signals.
Time (top) and frequency (bottom) representation of body sound signal acquired with the proposed system. The graphs show a 50 second segment of the recorded audio data which includes an apnea phase.
4 Discussion and conclusion
The first step in the development of a reliable and comfortable system for sleep monitoring is presented in this paper. Our system is capable of capturing heartbeats, breathing, snoring, sleeping positions and movements of the volunteer. However, a future study should examine large, randomly selected samples of volunteers including patients suffering from OSAS. To validate the proposed system a comparison with a golden standard system is needed. For this purpose, a comparison with standard polysomnography methods inside the sleep laboratory of the University Medical Center Ulm with a large number of patients is planned in the near future. The next step to improve our system is the development of a pattern recognition algorithm to automatically evaluate overnight recordings and deliver essential medical informations like the Apnea–hypopnea index.
Acknowledgment
The authors would like to thank Beurer GmbH for their assistance and support.
Funding: This study is part of the the project entitled ”SomnoSound” in cooperation with Beurer GmbH supported by the Arbeitsgemeinschaft industrieller Forschungsvereinigungen AiF (KF2186205AK3)
Author's Statement
Conflict of interest: Authors state no conflict of interest. Material and Methods: Informed consent: Informed consent has been obtained from all individuals included in this study. Ethical approval: The research related to human use has been complied with all the relevant national regulations, institutional policies and in accordance the tenets of the Helsinki Declaration, and has been approved by the authors’ institutional review board or equivalent committee.
References
[1] B. Stuck, Praxis der Schlafmedizin: Schlafstörungen bei Erwachsenen und Kindern; Diagnostik, Differentialdiagnostik und Therapie; mit 36 Tabellen. Berlin [u.a.]: Springer, 2. aufl ed., 2013.10.1007/978-3-642-34881-5Search in Google Scholar
[2] H. Pasterkamp, S. S. Kraman, and G. R. Wodicka, “Respiratory sounds. advances beyond the stethoscope,” Am. J. Respir. Crit. Care Med., vol. 156, pp. 974–87, Sep 1997.10.1164/ajrccm.156.3.9701115Search in Google Scholar PubMed
[3] H. Nakano, M. Hayashi, E. Ohshima, N. Nishikata, and T. Shinohara, “Validation of a new system of tracheal sound analysis for the diagnosis of sleep apnea-hypopnea syndrome.,” Sleep, vol. 27, pp. 951–7, Aug 2004.10.1093/sleep/27.5.951Search in Google Scholar PubMed
[4] A. Yadollahi, E. Giannouli, and Z. Moussavi, “Sleep apnea monitoring and diagnosis based on pulse oximetry and tracheal sound signals.,” Med Biol Eng Comput, vol. 48, pp. 1087–97, Nov 2010.10.1007/s11517-010-0674-2Search in Google Scholar PubMed
[5] C. Kalkbrenner, P. Stark, G. Kouemou, M.-E. Algorri, and R. Brucher, “Sleep monitoring using body sounds and motion tracking,” in Engineering in Medicine and Biology Society (EMBC), 2014 36th Annual International Conference of the IEEE, pp. 6941–6944, Aug 2014.10.1109/EMBC.2014.6945224Search in Google Scholar PubMed
[6] S. Madgwick, A. J. L. Harrison, and R. Vaidyanathan, “Estimation of imu and marg orientation using a gradient descent algorithm,” in Rehabilitation Robotics (ICORR), 2011 IEEE International Conference on, pp. 1–7, 2011.10.1109/ICORR.2011.5975346Search in Google Scholar PubMed
[7] B. Kemp, A. Värri, A. C. Rosa, K. D. Nielsen, and J. Gade, “A simple format for exchange of digitized polygraphic recordings,” Electroencephalogr Clin Neurophysiol, vol. 82, pp. 391–3, May 1992.10.1016/0013-4694(92)90009-7Search in Google Scholar PubMed
© 2015 by Walter de Gruyter GmbH, Berlin/Boston
This article is distributed under the terms of the Creative Commons Attribution Non-Commercial License, which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
