11 international conference on development and application

11th International Conference on DEVELOPMENT AND APPLICATION SYSTEMS, Suceava, Romania, May 17-19, 2012

75

Abstract — In this paper is described the realization of a

cardiac arrhythmia monitoring system using wireless sensor networks. The proposed system is able to measure and transmit to a central monitoring station the heart rate (HR). The system can be used for long-time continuous patient monitoring, as medical assistance of a chronic condition, as part of a diagnostic procedure, or recovery from an acute event. The HR are continuously measured using a custom developed devices and then transferred to central monitoring station via a wireless sensor network (WSN). The central monitoring station runs a patient HR monitor application that receives the HR from WSN and activates the alarms when a heart rate arrhythmia is detected. A user-friendly Graphical User Interface was developed for the HR monitor application to display the received measurements from the monitored patients. A prototype of the system has been developed, implemented and tested.

Index Terms — cardiac arrhythmia, heart rate, low power, remote patient monitoring, wireless sensor networks

I. INTRODUCTION

Due to the increasing occurrence of sudden death events caused by cardiovascular diseases, there is a need to pro-vide a long-time continuous patient monitoring services. The patient monitoring may be performed at a variety of environment, within hospital or their home, as medical assistance of a chronic condition, as part of a diagnostic procedure, or recovery from an acute event.

Heart rate (HR) is a physiological parameter commonly used by wireless patient monitoring systems. It allows an assessment of the condition of the patient, the cardiac arrhythmias can be recorded promptly and variations can be easily differentiated from normal/abnormal. This parameter was frequently used in studies and research projects, providing vital information on the cardiovascular function.

Wireless monitoring represents a medical practice that involves remotely monitoring patients who are not at the same location as the health care provider [1]. Generally, a patient have a number of monitoring devices at home, and the results of these devices will be transmitted to the central monitoring station.

The homecare monitoring of patients with chronic diseases or elderly also represents an alternative to medical supervision within hospitals [2].

In the last years, the steady advances of the integrated

This work was supported by the project PERFORM-ERA "Postdoctoral Performance for Integration in the European Research Area" (ID-57649), financed by the European Social Fund and the Romanian Government

circuits technology, wireless networks, and medical sensors have opened the way to miniature, low power, and intelligent monitoring devices, suitable for many portable medical applications. Wearable heart monitoring devices allows to continuous monitor changes in HR and provides feedback to help maintain an optimal heart status. Long-time heart monitoring devices can provide information about the HR variations, useful as a recovery indicator in cardiac patients after myocardial infarction or help monitor effects of drug therapy.

Nowadays, there is a significant increase in the number of various wearable heart monitoring devices, ranging from simple pulse monitors and HR monitors, to portable Holter monitors. Although Holter monitors are used only to collect data, it still remains the most used device. Data processing and analysis are performed offline, making the device impractical for continual monitoring and early detection of heart diseases. Although widely used, they are not suitable for long time monitoring due to their limited power supply.

Monitoring patient’s HR within hospital or his home requires the use of sensors attached by wires to the medical devices, which limits the patient's activity. As an alternative, wireless devices are suitable for remote patient monitoring, giving him the freedom of movement.

The HR is continuously measured by the proposed system using custom developed devices. The results may be wirelessly transmitted to central monitoring station by using Bluetooth or WiFi nodes, but they are more expensive, consume more power, require installing an expensive infrastructure, and are useful for high bandwidth applications. Another Bluetooth limitation is that the standard allows only a limited number of nodes. These issues make WiFi and Bluetooth nodes unsuitable for widespread wireless monitoring of patient’s HR. As an alternative, wireless sensor networks, containing compact sensor nodes and having low power consumption, represent a cost-effective solution. The ZigBee is similar to Bluetooth but is simpler, has a lower data rate, and less power consumption, making it suitable for indoor applications.

This paper describes a system based on wireless sensor nodes for patient monitoring within hospital or his home. The sensor nodes contain custom developed devices that perform the measurements and transmit the results to a central monitoring station. The central monitoring station runs a patient HR monitor application that displays the results and activates the alarms when a heart rate arrhythmia is detected.

Remote Cardiac Arrhythmia Monitoring System Using Wireless Sensor Networks

Cristian Rotariu, Vasile Manta, and Razvan Ciobotariu “GHEORGHE ASACHI” TECHNICAL UNIVERSITY OF IASI

Dimitrie Mangeron no. 27, [email protected]


76

II. MATERIALS AND METHODS

A conceptual view of the proposed system consists of the following components: a wireless sensor network (WSN) used to measure HR from the patient, each sensor node has a ECG based HR detector attached on the patient, several repeater nodes distributed in WSN at fixed location, their number and density depending by the coverage requirements, and a central monitoring station running a patient HR monitor application. The application receives the HR from WSN, displays it as temporal waveforms, and activates the alarms when a heart rate arrhythmia is detected.

The overall architecture of the proposed system is represented in Fig.1.

Fig. 1. Cardiac Arrhythmia Monitoring System

Each sensor node contains a custom developed ECG amplifier connected to an eZ430RF2500 module from Texas Instruments [3], as it is represented in the Fig.2.

Fig. 2. ECG amplifier connected to eZ430RF2500 module

The 3-lead ECG amplifier is custom made device based on low power instrumentation amplifiers, has for each channel a high gain (x500), is AC coupled, and has a frequency band limited to 35 Hz. The high common mode rejection (>90dB), high input impedance (>10MΩ), the fully floating patient inputs are other features of the ECG amplifier. The amplifier is described in detail in [4].

The eZ430RF2500 module is a small wireless radio development kit from Texas Instruments based on the MSP430F2274 microcontroller [5] and CC2500 wireless transceiver [6]. The eZ430RF2500 module has a limited communication range (approx. 10m) and necessitates repeaters to send the measured data to central monitoring station.

The low power consumption of sensor nodes is an

important characteristic of the WSNs and contributes not only to prolonged lifetime of the sensor nodes, but also to the system miniaturization. For a sensor node the overall power requirements are represented by the sum of power requirements of each component. Therefore we also chose a low-cost 2.4GHz transceiver (CC2500) designed for very low-power wireless applications, circuit is intended for the 2400–2483.5 MHz ISM (Industrial, Scientific and Medical) and SRD (Short Range Device) frequency band. The transceiver consumes less than 21.2mA in transmission mode at 0dBm output power and 17.0mA in receiving mode.

The power supply of the sensor node is provided by two 1.2V rechargeable batteries through an inductorless DC/DC step-up converter – TPS60204 from Texas Instruments [7]. The converter generates a 3.3V±4% output voltage from a 1.8V to 3.6V input voltage, is typically powered by two alkaline, NiCd, or NiMH battery cells, and operates down to a minimum supply voltage of 1.6V.

The prototype of the sensor node is represented in Fig. 3.

Fig. 3. Sensor node

We decided to use the SimpliciTi protocol from Texas Instruments to transfer data from sensor node to central monitoring station [8]. SimpliciTI has as the main features low memory needs, advanced network control, and sleeping modes support. It is intended to support the development of wireless networks containing battery operated nodes and require low data rates.

The eZ430RF2500 module connected to the ECG amplifier was configured as End Device (ED), the eZ430RF2500 connected to central monitoring station as Access Point (AP), and several others are configured as Range Extenders (RE). Data transmission rate between the ED and AP through RE depends on patient’s HR, and usually range from 0.5Hz (for HR = 30bpm) to 3Hz (for HR = 180bpm).

The flowchart of the software working on MSP430F2274 microcontroller from the ED is represented in the Fig. 4. In this instance, the eZ430RF2500 module initialize onto the network, then, after a START command, wakes up to read the ECG signals. In order to compute the HR, the ECG signals are sampled at 200Hz with 10bits/sample. Also, the MSP430F2274 reads the battery voltage, runs a HR detection algorithm, and communicates the results to central monitoring station through RE and AP. In order to minimize the energy waste, since an important power consumer element is the radio transceiver, the CC2500 entered into low power mode after each transmission cycle.


77

Link toAccess Point

Read data fromAccess Point

Enter low power mode and wait for

timer interrupt

Acquire ECG samples and

battery voltage

Compute Heart Rate

Format data for transmission

Send data to Access Point

Read data fromAccess Point

STARTCommand?

Initializeradio

STOPcommand?

Timerinterrupt

occurred?

YES

NO

NO YESNO YES

Fig. 4. Flowchart of the software working on the MSP430F2274.

A user-friendly Graphical User Interface (GUI) was developed for the patient HR monitor application, to display the received measurements and alarms from all monitored patients. The GUI running on the central monitoring station was developed by using LabWindows/CVI version 2009, and is represented in the Fig. 5. On the GUI, temporal waveform of HR signal for selected patient are displayed, and the status of sensor node (the battery voltage and distance from the nearby RE or AP). The distance is represented in percent computed based on received signal strength indication measured on the power present in the received radio signal (RSSI).

Fig. 5. Central monitoring station GUI

The monitored patient has a name previously entered and information from his medical record (limits above the alarms become active) is used by the alert detection algorithm. The following physiological conditions that may cause alerts are: sinus bradycardia if HR < 40bpm, sinus tachycardia if HR > 120bpm, sinus arrhythmia if ΔHR/HR over last 5 min. > 20%, HR variability if max HR variability > 10% over the last 15 heart beats. Also, the low battery

voltage if VBAT < 1.9V and low value for RSSI if measured RSSI < 30% may generate alerts.

The process of detections the heart beats constitutes a significant part of the most ECG analysis systems. In applications were rhythm detection is performed, only the location of the R wave is required. In other applications it is necessary to find and recognize the features of the ECG signal, such as the P and T waves, or the ST segment, for the automated classification and diagnosis. Many algorithms for the extraction of the ECG features based on the digital filters have been reported in the literature [9], especiallyalgorithms for the QRS complex recognition.

In order to compute the HR, the algorithm implemented on the software running on MSP430F2274 microcontroller detects the QRS complexes from ECG signal. The framework for QRS detection algorithm was derived fromthe Pan-Tompkins algorithm [10] and it is represented in the Fig. 6. The classification algorithm used by the QRS detector is detailed in [11] and [12].

Fig. 6. Pan-Tompkins QRS detector

III. RESULTS

The prototype of the system described above has been implemented and tested.

The accuracy of measurements for HR test was performed by using the METRON 430 patient simulator (Fig 7.). The simulator has following technical specifications: selectable HR from 30, 40, 60, 80, 100, 120, 140, 160, 180, 200, 220, 240, 260, 280 or 300bpm with ± 1% accuracy and output ECG signal amplitude from 0.5mV, 1mV, 1.5mV or 2mV with ± 2% accuracy. The simulator has also the possibility to generate ECG signals corresponding to the several common heart arrhythmias.

Fig.7. Sensor Node test hardware

Fig. 8 summarizes the measured values from simulated HR in the range of 30 – 300bpm. From Fig. 8 we can notice that highest HR measured by the system is above 280bpm.

The simulated HR was then forwarded through the WSN (configured as a sensor network with 1 AP and 3 RE) to the


78

central monitoring station. The GUI running on the central monitoring station displays the data correctly. Finally, by using the same simulator, we tested the arrhythmia detection algorithm.

Fig. 8. Measured results for different simulated HR

In order to analyse the current profile of the sensor node, the hardware used is represented in the Fig 9.

Fig. 9. Test hardware for sensor node current consumption

The largest contributor to current consumption is the CC2500 wireless transceiver. To calculate the average current consumption for the application running on MSP430F2274, we used the Rigol DS5022 oscilloscope to acquire the current profile and Matlab R2010 to compute the integral of the voltage curve – resulted an area under the curve of 1150µV*s (Fig. 10). In this way, we obtain:average_current = (measured_voltage/10Ω)/period_of_trans. = 1150µV*s / 10 Ω / 0.1 s = 0.115mA.

Fig. 10. Actual profile consumption for Node (one sample)

To calculate the battery life expectancy of a sensor nodeand assuming that two AAA rechargeable batteries still maintain a 1000mA*hr rating under the hypothetical

condition in which the batteries hold their voltage ideally and until their capacity is exhausted, we obtain:hours_of_operation = current_rating / average_current = 1000 [mA*hrs] / 0.115 [mA] = 8695 [hrs] / 24 [hrs/day] = 362 [days].

Fig. 11. Actual profile consumption for Node (10 samples)

IV. CONCLUSIONS

A prototype of cardiac arrhythmia monitoring system has been developed, implemented and tested.

The proposed system allows monitoring HR from a remote location without requiring the physician to take the measurements.

Remote monitoring of patients with wireless devices, preventive or after major medical events became a usual procedure in medical practice.

The proposed system allows persons with chronic diseases or elderly people to be monitored within their home, as an alternative to medical supervision in hospitals.

REFERENCES[1] A. Milenkovic, C. Otto, E. Jovanov, “Wireless sensor networks for

personal health monitoring: Issues and an implementation”, Computer Communications (Special issue: Wireless Sensor Networks: Performance, Reliability, Security, and Beyond, Vol. 29, pp. 2521-2533, 2006.

[2] C. Rotariu, H. Costin, et al., “An Integrated System for Wireless Monitoring of Chronic Patients and Elderly People”, in Proceedings of the 15th International Conference on System Theory, Control and Computing, ISBN:978-973-621-323-6, pp.527-530, 2011.

[3] eZ430RF2500 Development Tool User's Guide, MSP430 Wireless Development Tool, http://www.ti.com/tool/ez430-rf2500

[4] C. Rotariu, H. Costin, D. Arotaritei, and B. Dionisie, “A Wireless ECG Module for Personal Area Network”, Buletinul Institutului Politehnic din Iaşi, Tome LIV (LVIII), Fasc.1, (2008), Automatic Control and Computer Science Section, pp. 45-54, 2008.

[5] MSP430, http://www.ti.com/lit/ds/symlink/msp430f2274.pdf[6] CC2500, http://www.ti.com/lit/ds/symlink/cc2500.pdf.[7] TPS60204, http://www.ti.com/product/tps60204.[8] SimpliciTI, http://www.ti.com/tool/simpliciti[9] B. Kohler, C. Hennig and R. Orglmeister, “The Principles of software

QRS Detection – reviewing and Comparing Algorithm for Detecting this Important ECG Waveform”, IEEE Eng. în Medicine and Biology, Jan-Feb, 2002.

[10] J. Pan and W.J. Tompkins, “A real-time QRS detection algorithm,” IEEE Trans. Biomed. Eng., vol. BME-32, pp. 230-236, 1985.

[11] P.S. Hamilton and W.J. Tompkins, “Quantitative investigation of QRS detection rules using the MIT/BIH arrhythmia database,” IEEE Trans. Biomed Eng., vol. BME-33, pp. 1157-1165, 1986.

[12] P.S. Hamilton, Open Source ECG Analysis Software, E. P. Limited, Somerville, Mass, USA, 2002, http://www.eplimited.com/osea13.pdf.

11 international conference on development and application

Documents