Wearable patch-type ECG using ubiquitous wireless sensor network ...

14 downloads 3601 Views 4MB Size Report
Nov 26, 2009 - Sensor Network for Healthcare Monitoring Application. Hsein-Ping Kew. Department of Ubiquitous IT Engineering,. Graduate School of Design ...
Wearable Patch-type ECG using Ubiquitous Wireless Sensor Network for Healthcare Monitoring Application Hsein-Ping Kew

Do-Un Jeong

Department of Ubiquitous IT Engineering, Graduate School of Design & IT, Dongseo University, Busan, South Korea. +82-51-320-1771

Division of Computer & Information Engineering, Dongseo University, Busan, South Korea. +82-51-320-1771

[email protected]

[email protected]

ABSTRACT

1. INTRODUCTION

Nowadays, the technological advances in sensors network, integrated circuit and wireless communication have enabled the design of lightweight and low-cost sensor nodes particularly for healthcare application. Ubiquitous sensor network give new possibilities for monitoring of human biomedical signal using wearable sensors node and allow patients the freedom to move around but still under continuously monitoring. In this paper, a new concept for wearable patch-type ECG sensor node transmitting signal via an ultra low power consumption wireless data communications unit to personal computer using Zigbeecompatible wireless sensor node. The measured ECG signals carry a lot of clinical information for a cardiologist especially R-peak detection in ECG signals. R-peak detection generally uses the threshold value which is fixed. There will be error in peak detection when the baseline changes due to motion artifacts and signal size changes. Therefore, variable threshold method is used to detect the R-peak which is more accurate and efficient. In order to evaluate the performance analysis, R-peak detection using MITBIH databases and Long Term Real-Time ECG is performed in this research.

The healthcare industry is confronting a number of challenges; including skyrocketing costs, a growing incidence of medical errors, inadequate staffing, and lack of converge in rural and underserved urban areas. Simultaneously population aging has become one of the most significant demographic processes of modern times. An inevitable consequence of the demographic transition and the shift to lower fertility and reduced mortality, the ageing of the world’s population has many countries facing unprecedented numbers and proportions of older persons. Healthcare workers are under increasing pressure to provide better services to more people using limited financial and human resources. This paper proposed a solution to the current ubiquitous healthcare. The wide scale deployment of wireless networks, wearable computing will improve communication among patients, physicians and other medical healthcare workers as well as enable the delivery of accurate medical information anywhere anytime, thereby improving access and reducing errors. Recent advances in sensor technology allow continuous, real-time ambulatory monitoring of multiple patient physiological signals including electrocardiogram (ECG), body temperature, oxygen levels, glucose levels, respiration, blood pressure and etc. For better treatment purpose especially for elderly person at home, technologies advanced for ubiquitous healthcare monitoring is essential. Advances in ubiquitous wireless sensor networking have opened up new opportunities in healthcare monitoring systems. The future will see the integration of the existing specialized medical technology with pervasive and wireless network.

Keywords R-peak detection, ubiquitous, wearable patch-type ECG, Zigbeecompatible wireless sensor node

This paper presents a wearable patch-type ECG where it capable of recording and analyzing continuous ECG data received from the human body. The overall system architecture provides an application for recording activities, events and potentially important medical symptoms. The hardware allows data to be transmitted wirelessly from wearable patch-type ECG sensor node to a base station attached to server PC using IEEE802.15.4 Zigbee for signal processing. If any abnormality occurs at server, then there will be alarm alert system to alert the doctors’ PDA [1].

"Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.

A concept is proposed for wireless and wearable patch-type ECG sensor transmitting signal to a diagnostic station at the hospital. This concept able to follow up critical patients from their home while they are carrying out daily activities. In this paper, the wearable patch-type ECG monitoring system has several advantages compared to existing solutions in market. One of the

ICIS 2009, November 24-26, 2009 Seoul, Korea Copyright © 2009 ACM 978-1-60558-710-3/09/11... $10.00"

624

main advantages is it is easy to use, user friendly and no extra technical skills required. Besides that, the ECG sensor is easy to be replaced by patient himself as the system is a compact electronic electrode.

In biomedical signal analysis, processes involved in computeraided diagnosis and therapy is shown in Figure 1. Biomedical signal for example ECG are obtain from patient through biomedical instrumentation. Thus, all biomedical instrumentation and signal analysis systems should be design for patient convenient and do not cause any danger or harm. In signal data acquisition portion have transducer, signal-conditioning equipment and analog-to-digital conversion. Sensor and ECG patch-type electrode are example of transducer. Amplifier and filters are example of signal-conditioning equipment where it required increasing or decreasing the amplitude of a signal. Transducer does not amplify power.

With this new solution, wearable patch-type ECG monitoring system brings much more convenient to patients where only one ECG transmitter is required for the ECG recording. The electrode is equipped with a wireless transmitter and battery supply for several days even with continuous usage. With the use of wearable patch-type ECG monitoring system, it is easier and more cost effective as compare to existing solutions in the current market. Patient able to continuously monitor at his/her home while doing daily activities. This paper describes the implementation of wearable patch-type ECG monitoring system using wireless sensor network technology [2], [3].

After the signal data is acquisitive, signal processing is next step. Filtering for removal of interference, artifacts or simply noise is important as most biomedical signal is appear as weak signals in general environment due to its teeming with other signal from various origin. In this step, selecting an appropriate filter such as time-domain filters, frequency-domain filters, and adaptive filters is taken. Event and wave detection is important to identify the part of the signal related to a specific event of interest. The P, QRS, and T waves in ECG can be determined.

The ECG features are used to detect life-threatening arrhythmias, with an emphasis on the software for analyzing the P-wave, QRS complex and T-wave in ECG signals at server after receiving data from the base station. This paper also analysis the ECG signals using variable threshold method for R-peak detection.

Once the events are determined, signal analysis is required for the next step. Analysis of non-stationary signals is carrying out either using fixed segmentation or adaptive segmentation. Pattern recognition based upon signal analysis is needed to classify of a given signal into one of many categories and further assist the diagnostic procedure. The final steps or purpose of biomedical signal analysis is to diagnostic decision corresponding to the condition of the patient.

2. SYSTEM DESIGN 2.1 Biomedical Signal Analysis Biomedical signal analysis is important in term of information gathering, diagnosis, monitoring, therapy and control, evaluation. Information gathering need to measure the phenomena to interpret a system; diagnosis is to detect any malfunction, abnormality; monitoring is to continuously gather information about the system; therapy and control need to modify the behavior of a system based upon the outcome to meet the requirements; evaluation is last step where check the ability to meet functional requirements, obtain proof of performance, quality control.

2.2 ECG Signal Human body is made up of many component systems, for example cardiovascular system, musculoskeletal system, nervous system, etc. Each system is further made up of several subsystems, for example cardiovascular system function is to pump blood through the pulmonary system for oxygenation and delivery of nutrients. It carries many physiological processes. These processes manifest themselves as signals in many types like biochemical, electrical, physical, etc. A pathological process associated with signal that can cause alterations in disease or defects. One of the simple examples of a biomedical signal is body temperature. These biomedical signals are important in assessment of the critically ill patients or a child regardless of its simplicity. There are various types of biomedical signals and each signal is useful in diagnosis. For example, ECG, EMG (electromyogram), VMG (vibromyogram), speech signal and etc. Among the most commonly known biomedical signals are ECG signals. ECG signals can be recorder from the surface electrodes on the limbs or chest. ECG is an electrical manifestation of the contractile activity of the heart. Human heart have the rhythm in bpm (beats per minutes), it can easily compare or estimated by readily identifiable waves. ECG is important in analysis cardiovascular diseases, ventricular hypertrophy, and myocardial ischemia because it can be altering using ECG wave shape. Thus, diagnosis decision regarding condition of patient can be made by physician or medical specialist by altering ECG wave shape.

Figure 1. Biomedical signal analysis based on computer-aided diagnosis and therapy.

625

Figure 2. Composition of typical ECG signal recorded from body surface. Figure 3. Overall ECG monitoring system architecture.

The science of measurement of physiological variables and the parameter is called biometrics. The heart normal produce regular rhythmic, arrhythmia is the disturbance in the regular rhythmic activity of the heart. In general, a typical ECG signal tracing of a normal heartbeat or a normal cardiac cycle consists of a P wave, a QRS complex and a T wave as shown in Figure 2. ECG can be recorded by placing electrodes placed on the surface of the body.

This paper discusses to realize an ambulatory ECG measurement system performed in medical institutions. Figure 3 illustrates the overall system architecture of ECG monitoring system. An ECG analysis with activity monitoring for the home care of elderly person or patients, using wireless sensor technology was design and implemented. Patient is wearing a wearable patch-type ECG electrode on his/her chest to measure the ECG signals. An ECG transmitter sensor node is attached to the wearable patch-type ECG electrode. At the end user which is PC, there will be a ECG receiver to receive the signal transmitted via wireless transmission Zigbee 2.4GHz. ECG monitoring program is able to display at the PC and further analyze the signal can be done easily.

Event and wave detection is important to identify the part of the signal related to a specific event of interest. The P wave, QRS complex, and T waves in ECG signal can be determined. The approximate values for the durations of various waves and intervals of a healthy adult heart are shown in Table 1. However, it may vary depend on the age or gender of heart rate.

Table 1. Duration of ECG Parameters in a normal adult heart ECG Parameter

Duration (second)

P Wave

0.08-0.10

QRS Complex

0.06-0.10

P-R Interval

0.12-0.20

Q-T Interval

0.30-0.40

Long term ECG monitoring and analyzing plays an important role in heart disease and high-risk cardiac patients. ECG event classification is the main objective to further enhance medical treatments. Thus, a more effective and reliable extraction of characteristic ECG parameter is needed to classify the ECG signal.

Figure 4. Overall figure of wearable patch-type ECG electrode.

3. ECG MEASUREMENT SYSTEM Ubiquitous healthcare component consists of sensing, monitoring, analyzing, disease classification and emergency alert. In healthcare monitoring system, emphasis is placed on sensing, monitoring and analyzing [4], [5], [6].

A wearable patch-type ECG with integrated electronics has been developed and has proven to be long-term robustness of all electrical components. The wearable patch-type ECG has been developed with three integrated dry patch electrodes.

626

Figure 7. Block diagram of analog signal processing. ECG measurement system at the PC monitoring converts the analog ECG signal which is detected using TIP710CM (Maxfor, Co., Korea) and an ultra-low-power Zigbee-compatible wireless transmission sensor node. The sensor node was designed based on TI’s low power consumption MSP430F1611 microprocessor, build in 12-bits Analog Digital Converter which sampled the analog ECG signal at a rate of 500 times per second. It also applied to the sensor node with Zigbee CC2420 radio chip which has a build in short-range wireless sensor network.

Figure 5. Wearable patch-type ECG electrode. The ECG signal which is measured at the surface of the skin using electrode is approximately 1mV. The signal measured is not only from ECG signal as a parts of the muscle tissue that occur in the excited biological signal, but also from stray capacitance of the surrounding environment and various electronic equipments from the unwanted noise signal. Therefore, ECG signal processing circuit is designed to amplify the signal and to extracts only the desired ECG signal.

On the PC side, another sensor node which is the ECG receiver, receive the signal transmitted by the radio. There is a PC Monitoring Program to display the ECG data and save the required ECG data for a specified time. Visual Studio 2006 is used for the real-time display monitoring program in PC. The actual implementation of this research is wearable patch-type ECG electrode that wear around the patients’ chest was designed as shown in Figure 8.

Firstly, from the displacement amplification, it undergoes lowpower amplifier (INA326, Texas Instruments Co., USA). In order to remove power line noise, a 60 Hz TwinT Notch filter was designed for variable of Q value. TwinT Notch filter is useful in rejecting unwanted signals that are on a particular frequency. The filter’s response consists of a low level of attenuation which is away from the notch frequency. As the input signal move closer to notch frequency, attenuation level will increase this giving the typical notch filter’s response.

A preamp component which consists of signal amplifier is used to detect the ECG signal induced from wearable patch-type ECG electrode. Preprocessing method is used to obtain information about signal slope and intensify frequency response curve of derivative.

Figure 8. Data Processing on receiver. Firstly, differentiation process is performing on the ECG signal. By applying the window in differential ECG, the maximum amount of points with the slope, the derivative procedure to determine the R-peak.

Figure 6. Wearable patch-type ECG transmitter sensor node. In order to minimize the removal and basis line fluctuation, gain control and buffer are used in signal amplification circuit. A secondary 35 Hz low pass filter was used at the final analog ECG signals. The block diagram of configuration ECG measurement circuit for analog signal processing is shown in Figure 7.

In differential part, the derivative operation is specified below is used. 1

y(n) = [2x(n) + x(n − 1) − x(n − 3) − 2x(n − 4)] 8

627

(1)

Differentiating the ECG signal for the purpose of modifies its phase, creating zero-crossing in the location of R-peaks. The derivative-based operator is used to remove base-line drift in the ECG signal and low-frequency artifacts [7].

Firstly, MIT-BIH Long Term ECG Database is used. Second part will be the Long Term Real-time ECG analysis.

4.3 R-peak Detection using MIT-BIH Long Term ECG Databases

After the pre-processing methods, variable threshold method is used to further detect the R-peak. The formula for variable threshold value is defined as

VTH =

∑tt≡t−4 (Rpeak t )−Max (Rpeak ) 4

× 55%

For performance analysis purpose, R-peak detection with variable threshold method using MIT-BIH Long Term ECG Database 14184 was utilized. The database is a long-term ECG recording with time range of 14 to 22 hours and manually reviewed beat annotations.

(2)

After applied signal preprocessing technique on MIT-BIH Long Term ECG Database 14184, the result is shown in Figure 10(a). Then, by using a fixed baseline value and threshold value method, the R-peak detection is shown in Figure 10(b). The red dotted line shows the fixed threshold value and red color circle show the Rpeak detection. In the larger screenshot with the range of 30 s to 40 s, there are total of 5 out of 15 R-peak is not successfully detected as show in Figure 10(b) with a black color cross. In the experiment result, it shown that the R-peak is not found and detected at the level below the fixed threshold value which is 3 V in amplitude.

To apply the baseline value, value corresponding to 55% of data average 4 except a highest from 5 R-peak detection from ECG is detected variably first. The algorithm was implemented in MATLAB.

4. EXPERIMENTS AND RESULTS 4.1 ECG Preprocessing Results In preprocessing part, the input ECG signal undergoes differentiation process, then Hilbert transform. The experimental performance of the ECG preprocessing is performed. Firstly, ECG signals of MIT-BIH Arrhythmia Database 100 in the range of 0 to 10 second are used as shown in Figure 9(a). Then, the signal undergoes differentiation process as shown in Figure 9(b).

In this research, variable threshold value technique is proposed in order to solve this problem. As shown in Figure 10(c), all the Rpeak of ECG is successfully detected. It detected the R-peak using variable threshold value method which it previously not detected.

(a) Original MIT-BIH 100 ECG (a) MIT-BIH Long Term ECG Database 14184

(b) After differentiation process Figure 9. ECG signal preprocessing.

4.2 R-peak Detection Performance Analysis In R-peak detection of ECG signal, it generally used the threshold value which is fixed. However, it does not give accurate results in R-peak detection when the baseline changes due to signal size and motion artifacts.

(b) Fixed Threshold Value

In this research, a variable threshold value is used in order to overcome the problem and to have a more accurate R-peak detection results. To apply the threshold value which corresponding to 55% of the data average, 4 out of the 5 R-peak of ECG is firstly detected. Two part of performance analysis are evaluated.

628

As discussed in previous part, variable threshold give a more accurate results as compare to fixed variable value. Thus, the end result of the R-peak detection used variable threshold value. After the preprocessing method and variable threshold method, Rpeak is detected in the signal. From the R-peak, R-R interval can be calculated easily as shown in Figure 12. The R-R interval is also known as inter-beat intervals. R-R interval is useful in providing information of HRV (Heart Rate Variability). HRV is the measurement of beat-to-beat variations in heart rate and is useful in analyzing cardiovascular autonomic control [10].

(c) Variable threshold value method Figure 10. R-peak detection results using MIT-BIH. Figure 12. R-R Interval using real-time ECG.

4.4 R-peak Detection using Real Time ECG In this part, R-peak detection is detected using Long Term Real Time ECG and the same signal processing algorithm process. Figure 11(a) show the original of Long Term Real Time ECG from the range of 0 to 1800 seconds.

5. CONCLUSION A wearable patch-type ECG electrode with activity monitoring system was developed for the advanced ubiquitous healthcare system using sensors technologies. The wearable patch-type electrode able to transmits ECG signal to PC monitoring via wirelessly and a Zigbee-compatible wireless sensor node.

Figure 11(b) show the R-peak detection result after preprocessing method which include differentiation, Hilbert transform and using variable threshold value. After undergoes preprocessing method, the amplitude of the signal is adjusted. The red color circle shown that the R peak is detected and give a good result where all the peak is detected [8], [9].

A signal processing algorithm was developed for R-peak detection using real time processing procedure. Signal processing algorithm is implemented in MATLAB where preprocessing process is the first step which includes differentiation. In order to detect the Rpeak of ECG, variable threshold is purposed to replace conventional fixed threshold value. The performance of fixed and variable threshold using MIT-BIH long terms ECG database and long term real time ECG are evaluated. It shows that variable threshold method used has improved the accuracy in R-peak detection. From the R-peak, R-R interval can be easily obtain and is useful information for HRV as HRV is one of the most promising value for cardiologist. Ubiquitous healthcare has the potential to reduce long-term costs and improve quality of service, but it also faces many administrative and technical obstacles. Thus, in future, more research on the wireless sensor network technology for ubiquitous healthcare system is needed.

(a) Original Long Term Real Time ECG

6. ACKNOWLEDGMENTS This research was supported by Basic Science Research Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Education, Science and Technology (20090074866)

7. REFERENCES [1] K. Y. Kong, C. Y. Ng, and K. Ong, “Web-Based Monitoring of Real-Time ECG Data,” Computers in Cardiology, vol. 27, pp. 189-192, 2000. [2] D. Konstantas, “The Mobihealth Project. IST Project,” IST2001-36006, European Commission: Deliverable 2.6, http://www.mobihealth.org, 2004.

(b) R-peak Detection Results Figure 11. Long term real time ECG.

629

[3] E. Jovanov, D. Raskovic, J. Price, J. Chapman, A. Moore, and A. Krishnamurthy, “Patient Monitoring Using Personal Area Networks of Wireless Intelligent Sensors,” Biomed Sci Instrum, vol. 37, pp. 373-378, 2001.

Transaction on Biomedical Engineering, Vol. 55, No. 2, February 2008. [8] Do-Un Jeong, Hsein-Ping Kew, “Real-Time Monitoring of Ubiquitous Wearable ECG Sensor Node for Healthcare Application,” ICCSA 2009, Part 1, LNCS, vol. 5592, pp. 868884, Springer, Heidelberg, 2009.

[4] Victor Shnayder, Bor-rong Chen, Konrad Lorincz, Thaddeus R. F. Fulford-Jones, and Matt Welsh, “Sensor Networks for Medical Care,” Harvard University Tehnical Report TR-0805, April 2005. [5] Lim L. and B. Yee, “Coach’s Companion – Athlete’s Health Monitoring System,” University of California, Berkeley, 2005.

[9] Do-Un Jeong, Se-Jin Kim, “Development of a Technique for Cancelling Motion Artifact in Ambulatory ECG Monitoring System,” Proceedings of the 2008 Third International Conference on Convergence and Hybrid Information Technology, vol. 1, pp. 954-961, November 2008.

[6] Rune Fensli, Einar Gunnarson, Torstein Gundersen, “A Wearable ECG-recoding System for Continuous Arrhythmia Monitoring in a Wireless Tele-Home-Care Situation,” Proceedings of 18th IEEE Symposium on Computer-Based Medical System (CBMS’05), 2005.

[10] Task Force of The European Society of Cardiology and The North American Society of Pacing and Electrophysiology, “Heart rate variability, Standards of measurements, physiological interpretation, and clinical use,” European Heart Journal, vol. 17, pp. 354-381, March 1996

[7] Natalia M. Arzeno, Zhi-De Dang, Chi-Sang Poon, “Analysis of First-Derivative Based QRS Detection Algorithm,” IEEE

630