Remote continuous cardiac arrhythmias detection and monitoring.

Abstract

The current techniques used to diagnose cardiac arrhythmias such as Holter, Rtest and telemetry systems are partially efficient because they are limited either in time or in space. In this paper, a platform dedicated to the real-time remote continuous cardiac arrhythmias detection and monitoring is proposed. Such a platform allows to improve the accuracy and the efficiency of the diagnostic of ventricular tachycardia among the high-risk patients and enables the implantation of ICD to prevent sudden death. The new method allows the patient to lead a normal life while being remotely monitored in real-time by an ambulatory wireless ECG sensor. When a cardiac arrhythmia is detected a message including a sequence of ECG signals and the patient's images (indoors only) is sent to a remote surveillance server. According to the gravity of the symptom, the cardiologist can intervene in real time or later. The system has been evaluated on some ten patients with regard to heartbeat and cardiac rhythm disturbance. The real-time results are similar to those offered by HP telemetry systems.

Full text

Page 1

1
Remote Continuous Cardiac Arrhythmias
Detection and Monitoring
Haiying Zhou*, Kun Mean Hou*, Jean Ponsonnaille**, Laurent Gineste *, Julien Coudon*,
Gil de Sousa*, Christophe de Vaulx *, Jian-Jin Li* , Pierre Chainais*, Romuald Aufrère*,
Abdelaziz Amamra* and Jean-Pierre Chanet*
*Laboratoire LIMOS UMR 6158 CNRS, ISIMA, UBP Clermont-Ferrand II, France
**CHU de Clermont-Ferrand, Université d’Auvergne, France


Abstract. The current techniques used to diagnose cardiac arrhythmias such as
Holter, Rtest and telemetry system are partially efficient because they are limited
either in time or in space. In this paper, a platform dedicated to the real-time remote
continuous cardiac arrhythmias detection and monitoring is proposed. Such a
platform permits to improve the accuracy and the efficiency of the diagnostic of
ventricular tachycardia among the high risk of death patients and enables to propose
the implantation of ICD to prevent sudden death. The new method allows the patient
to lead a normal life while being remotely monitored in real-time by an ambulatory
wireless ECG sensor. When a cardiac arrhythmia is detected a message including a
sequence of ECG signal and the patient images (indoors only) are sent to the remote
surveillance server. According to the gravity of the symptom the cardiologist can
intervene in real-time or lately. The system is evaluated on about ten patients: heart
beat and cardiac rhythm disturbance. The real-time results are similar to the HP
telemetry system ones.

Keywords: Telemedicine, real-time remote continuous cardiac arrhythmias
detection, wireless ECG sensor network, remote monitoring server.
1. Introduction
Thanks to the rapid developments of pathological research and clinical technologies, most
of heart diseases can be effectively treated and prevented in today’s society. Nevertheless, it
can’t change the fact that heart disease is still the world’s number one killer. It is
responsible for one in every three deaths, which is an estimated seven million deaths in the
world each year [15].
Most of them are sudden cardiac death after heart attack. The sudden death is defined as
a death arising less than one hour after the first symptoms felt by the patient. It concerns
about 50,000 persons per year in France. 90% of the sudden deaths are due essentially to
the cardiac arrhythmias: 20% are caused by heart block or pause (bradycardia) and 80% are
caused by the ventricular fibrillation (VF), frequently initiated by the ventricular
tachycardia (VT). The principal aetiology of sudden death for adults is due to myocardial
infarction. For the group of population having a coronary pathology and chronic ischemia,
the risk of death is particularly high.
This "massive heart attack" is generally considered as an unpredictable and
unpreventable event. In spite of the effectiveness of the post-heart-attack treatment, the
patients die because heart attack usually occurs suddenly, without a shred of warning.

Page 2

2
Recent studies have shown that there are common significant cardiovascular abnormal
symptoms such as palpitations, faints, chest pain, shortness of breath etc, before the sudden
appearance of a lethal heart arrhythmia. If these symptoms can be early detected and
diagnosed, time is won to prevent the occurrence of heart attack. Therefore, to reduce the
number of disabilities and deaths caused by heart attack, it is necessary to have an effective
method for early detection and early treatment.
The most effective preventive therapy of sudden death due to cardiac arrhythmias is the
implantation of an implantable cardioverter-defibrillator (ICD). ICD is used to apply a
strong electrical shock to the heart. By adjusting cardiac rhythm to orderly and effective
status, this device helps to treat cardiac disorders such as ventricular fibrillation, ventricular
tachycardia, atrial fibrillation, and atrial flutter. Unfortunately, its high cost is a main factor
that impedes ICD to be widely applied. Moreover, it is an invasive technique requiring a
major surgery with potential complications. The complications that a physician may
encounter during surgery are the venous access, lead placement, intravascular
thrombosis/fibrosis, and the generator [16].
Currently, ICD is mainly applied to the high risk death patients who have cardiac
arrhythmia especially VT or VF, when the risk is accurately identified. Nevertheless, recent
surveys discover sudden cardiac death not only occurs in people who have had heart attacks
(myocardial infarction) in the past, but also can occur in young people who were entirely
well until they died [17]. Therefore, we need an effective personal diagnosis system which
can continuously monitor cardiac status. This system should be cost-effective, risk-free and
easy to use in daily life.
The ECG (Electrocardiograph) is the most commonly performed cardiac test, because it
is a useful screening tool for a variety of cardiac abnormalities; the test is simple to
perform, risk-free and inexpensive. From the ECG tracing, the following information can
be determined [18]:
? the heart rate
? the heart rhythm
? the conduction abnormalities: abnormalities in the way the electrical impulse spreads
across the heart
? the coronary artery disease
? the heart muscle abnormality etc.

The HOLTER technique is no doubt frequently used to record 24h or 48h of 2 leads
ECG signals. The recorded ECG signals will be analysed by a dedicated software and a
report is produced to be interpreted by the cardiologist, but it is proved largely insufficient
for a long term prediction because the critical cardiac arrhythmias do not necessarily occur
during the 24h or 48h [19]. Another technique called RTEST allows the patient to monitor
the record of a sequence of the ECG signals [11]. The recorded ECG signals may be sent to
a remote server or analyzed later. One of the drawbacks of the RTEST technique is that
most of the time the patient does not feel the palpitation or the VT. In fact, some cardiac
rhythm disturbances are just asymptomatic. The new generation of RTEST may be
configured by the physician to record automatically the ECG signals. Thus, these two
techniques are limited in time (4 weeks for the RTEST) and are proved insufficient and
partially efficient. Furthermore, the telemetry cardiac arrhythmia detection system (for
example Agilent) is expensive and limited in space because in general, it is only installed at
the cardiology department of a hospital.
Therefore, it is very important to propose a new method permitting to improve the
efficiency and the accuracy of the cardiac arrhythmia diagnostic which is not limited in
time and in space like the previous ones. Furthermore, the comfort of the patient has to be

Page 3

3
taken into account by allowing the patient to live normally and stay at home. All cardiac
rhythm disturbances will be recorded and analysed continuously and automatically in real-
time and, according to the gravity of the symptom, an emergency message may be sent to a
remote server. The cardiologist can confirm or reject the emergency message by analysing
the sequence of ECG signals included in the emergency message. If necessary, he can also
remotely analyse in real-time the patient ECG signals.
In fact, we believe that such a platform is able to improve the result of the prediction and
the diagnostic for the patient having cardiac arrhythmias, because it is not limited in time
and in space. Moreover, the patient feels easy and more secure at home. When a high risk
of death patient is definitely identified an ICD will be proposed. Finally, the platform is
able to monitor a large number of patients at home.
This paper is organized as follows. In section II the system elements and operation
modes of the platform are described. Section III presents the key techniques of the
platform. Finally in section IV, performance test and evolution, the conclusion and the
ongoing work are presented.
2. Overview of Platform
Our main objective is the development of a platform adapted to telemedicine applications
especially to real-time remote continuous cardiac arrhythmias detection and monitoring.
The platform integrates the advanced wireless telecommunication technology such as WiFi,
Bluetooth, GSM and UMTS and the distributed embedded real-time intelligent sensors
communicating over Internet.


Figure 1 - Platform dedicated to real-time remote continuous cardiac arrhythmias detection and monitoring
2.1 Platform elements
The platform structure shown in figure 1 comprises two parts: a local system and a remote
system, which contains 4 main configurable elements. The local system has a wireless ECG
sensor (WES) and a local server, and the remote system includes a remote server and a
remote surveillance system (Diagnosis and visualization).

Page 4

4

Figure 2.a - Wireless ECG Sensor Figure 2.b - Remote surveillance server
2.1.1 WES
To realise a low cost, low energy consuming and compact wireless ECG sensor (WES)
responding to the last AHA recommendations [6], the embedded basic technologies such as
distributed real-time fault tolerant micro kernel [3, 4], dedicated hardware and firmware
[12] and real-time TCP/IP protocol stack [1, 4] are implemented in the WES.
The wireless ECG sensor prototype (figure 2-a) is a real-time wireless embedded
portable sensor (size=70*100mm) based on the Texas Instruments ultra low power micro-
controller MSP430 [13] corresponding to the recommendations of AHA [6]. The sensor
without wireless adapter consumes only 10mA. The key features of the WES are:

? Gain: 1000
? CMMR(min): 120dB
? Bandwidth: 0.05Hz to 125Hz
? Programmable sample frequency more than 500Hz
? Analogue to digital converter: 12 bits
? Leakage current: 10µA.

The WES enables to capture 4 leads ECG signals sampled at 500Hz in real-time (The
sample frequency is reprogrammable). These sample signals are sent simultaneously to the
local server over wireless medium such as WiFi or Bluetooth. In off-line mode, ECG
signals will be stored into a flash memory of the WES. The duration of the ECG records
depends on the capacity of the flash memory card, the sample frequency and the number of
ECG leads. The two latter parameters may be configured by the user. For example, 128M
flash memory card can store 24h continuously 4 leads ECG signals sampled at 500Hz. In
this way, the WES works as a HOLTER or a RTEST.
2.1.2 Local server
The local access server that may be implemented by a standard PC or a mobile phone or a
dedicated network access medium, provides two network medium access services: wireless
connection with the WES by wireless medium (WiFi or Bluetooth), and network
connection with the remote system by multi-support network access mediums.
The different access mediums that patients can use are: Modem, broadband, wireless and
satellite connections. The network bandwidth of the access mediums is fluctuated in time
because it is affected by various disturbance factors [20]. Therefore, the remote system
must be adaptable to meet various network access medium bandwidths and local server
performance. So, in accordance with the network access medium and local server resources,
local peer can be configured to provide real-time ECG signals and patient‘s image
transmission and diagnosis. Otherwise, if the local server is a mobile phone, only short high
level alarm message is sent to the remote server. Thus, the local server may be configured
to support 4 different operation modes (section 3).

Page 5

5
For 4 leads ECG signals sampled at 500Hz, a 5 seconds frame contains 20,000 bytes of
ECG signal data (4 leads x 500Hz x 5s x 2 bytes). The fluctuation of 56Kbps MODEM
bandwidth does not allow real-time continuous transmission of ECG data. So, it is
important to minimize the amount of transmission data to reduce network traffic load. A no
loss ECG signal compression algorithm is implemented by taking into account the
resolution of ADC ‘Analogue to Digital Converter’ of the WES and the type of ECG
signals. The compression ratio can attain 50~60%, so only 8000~10,000 bytes of data are
transmitted. Furthermore, if ECG signals diagnosis is performed by the local server, only
25% of ECG signals raw data (2000~2500 bytes per frame of 5s) are really transmitted to
the remote server for its display. In fact, the ECG signals sampled at 125Hz are acceptable
for its visualisation.
Depending on the network traffic, the patient’s images captured from a webcam
connected to the local server are used to confirm the emergency state and remote diagnosis.
In fact, in spite of the advance of the techniques used to detect cardiac arrhythmias,
currently the accuracy of the results is around 90% [5, 7, 8, 9, 10]. Consequently, 10% of
emergency messages are false and it will be hard to manage the first aid. Therefore, for
real-time remote assistance and surveillance, the patient images are absolutely essential.
2.1.3 Remote server
The remote server provides network connection and patient database management. Thus
the remote server is composed of three servers: PPP server, WAP server and database
server. The PPP server allows the patient to connect to the remote server through a
traditional Public Switched Telephone Network (PSTN). The PPP server supports different
PSTN medium bandwidth: 56Kbps (standard Modem), 512Kbps and 1Mbps (ADSL). The
WAP provides seamless network connection over wireless mobile communication network
and it tunes automatically to the available medium bandwidth: GSM (9,6Kbps) and GPRS
(115Kbps). Moreover, in case of limited area such as a department of a hospital, the local
servers and the remote server may be configured to operate with the Ethernet LAN.
The database server stores patients’ ECG signals sequence, patient’ ECG diagnostic
reports, patients’ images, and patients’ profile and account information. Thus, at all times,
the physician can visualize the status of a patient and he can remotely reconfigure the
function mode of the local system.
2.1.4 Remote surveillance system
The remote surveillance system contains a visualization surveillance platform and a
background real-time communication system.
In order to improve the efficiency of data transmission, an adaptive communication
protocol with acknowledgment is implemented over UDP protocol (User Datagram
Protocol offering non-guaranteed datagram delivery) to deliver ECG signals. As we state
previously, ECG signals are compressed before sending to the remote system (each frame is
a window of 5 seconds ECG signals). So, the received ECG signals will be decompressed
and stored into data frame list and be displayed after 25~30 seconds delay (5~6 data
frames). The data buffering and delay are necessary to guaranty the real-time continuous
display of ECG signals.
Furthermore, compared with image data used to confirm diagnosis results, ECG data is
higher priority level of transmission. Thus, if the network traffic is heavy, the remote server
will request the patient peer (local server) to stop or reduce image transmission. Finally, in
order to guarantee the security of data transmission, a private key of 64 bits length is used
to perform encryption and decryption of all patient’s data.

Page 6

6
The interactive visualization surveillance platform (fig. 2b) enables to display
continuous ECG signals sequence and patients’ image, to respond to various alarm
messages, and to support real-time or on-line diagnosis. The 4 leads ECG signals and its
diagnosis results can be recorded into local data files in the format of WFDB [14]. It is to
be noted that the interactive visualization system provides the same GUI as today available
commercial devices (Agilent telemetry system, ELA etc.)
2.2 Operation modes
This platform enables 4 operation modes in order to adapt to different application
environments and requirements. The operation mode is decided by the physician by taking
into account the patient’s physical status and network medium access bandwidth. The key
features of the 4 operation are as follow:

1. Level 1: Real-time continuous ECG signal. For the sake of remote real-time displaying
and diagnosing, the data including continuous ECG signals acquisition and its detection
report will be sent in real-time to remote system. This operation mode is the highest
alarm level enables real-time on-line diagnosis. This mode is not appropriated to monitor
large number of patients due to the limitations of network bandwidth, system resources
and human resources but it is necessary to monitor high risk of sudden death patient. In
practice, each physician can survey approximately 4 patients. In this level, to assure
reliable cardiac arrhythmias diagnosis, patient’s image may be required.
2. Level 2: ECG signal sequence. In order to satisfy remote real-time multi-patients
detection and monitoring, the WES is configured to send automatically a sequence of
ECG signals (pre- and post- abnormality) to the remote system when a cardiac
arrhythmia event defined by the cardiologist is detected. This operation mode is suitable
for long-term multi-patients (lower risk of sudden death than the previous class of
patients) cardiac arrhythmias events surveillance.
3. Level 3: textual emergency message. In this mode, only a short textual emergency
message will be sent to physician when a cardiac arrhythmia event is detected.
According to the gravity of the symptom the physician can decide to intervene
immediately or later. This mode may be operated on any access medium (wire or
wireless).
4. Level 4: diagnosis report email. It’s the lowest level operation mode. The local server
will send periodically a report (like HOLTER report) attached to an email to the remote
server. The period is defined by the physician. This mode is suitable to monitor a large
number of patients.
It is to be noted that, the physician can remotely reconfigure the operation mode to adapt
to the evolution of the patient’s status.
3. Technological Overview
In terms of software development, the platform contains 4 main configurable modules. The
4 modules are configurable and loaded to implement an application tuning with system
resources to meet users’ requirements.
ECG acquisition module: The WES enables to capture 4 leads ECG signals sampled at
500Hz in real-time. The sample frequency and the lead numbers are programmable to meet
users’ requirements. The raw digital input ECG signals are filtered by a band pass (0.05Hz,
125Hz) and a notch (50Hz). Furthermore, in order to satisfy real-time multi-processes
operation, an adaptable embedded real-time micro kernel is integrated into the WES.

Page 7

7
ECG diagnosis module: Another key feature of this platform is real-time effective ECG
detection and diagnosis algorithm. This algorithm can automatically diagnose in real-time
tachycardia ventricular (TV), brachycardia ventricular (BV) and fibrillation ventricular
(FV) and other anomalies. Moreover, the algorithm is developed to ease its VLSI
implementation.
Embedded real-time communication module: The platform provides an embedded real-
time TCP/IP stack to supply network function for the WES. This minimization TCP/IP
stack contains essential real-time communication elements that support following
protocols: TCP, UDP/IP, ICMP and PPP. It also provides remote surveillance functions for
system management on SNMP standard and PING service.
Telemedicine communication module: A reliable and effective remote network
communication is the main foundation of telemedicine. The telemedicine communication
module provides a high layer adaptive communication protocol permitting to overcome
network access medium bandwidth fluctuation. Moreover, a compression algorithm is
implemented to reduce network traffic. Because of the different priority levels between
ECG and image data, a competition algorithm is designed to ensure real-time transmission
of ECG signals.
The telemedicine communication module ensures data reliability, network security and
peer to peer quality service.
4. Conclusion and ongoing work
Currently, the platform dedicated to the real-time remote continuous cardiac arrhythmias
detection is evaluated on about 10 patients at the C.H.U. of the Gabriel Montpied hospital
in Clermont-Ferrand (France). The detection algorithms were also evaluated by using MIT-
BIH data base [14]. Concerning VT and ESV, the detection rate is about 96% on the
patients. It is to be noted that, the quality of the ECG signal of our platform is better than
the HP telemetry system one.





Our remote
surveillance server
HP central display
screen
WES
HP ECG remote sensor
Figure 3 - Test and evaluation at the Gabriel Montpied’s hospital in Clermont-Ferrand
Our platform allows continuous remote cardiac arrhythmias monitoring and permits the
patient to lead a normal life indoors and outdoors thus it is more efficient to diagnose
cardiac arrhythmias.
The results of the test and evaluation until now are satisfactory. We believe that such a
platform enables a new clinical approach to evaluate more accurately a large number of the
high risk of sudden death patients. Furthermore, it may be used by the cardiologist to
remotely monitor and evaluate the efficiency of the drug or to discuss a difficult cardiac

Page 8

8
pathology with other colleagues.
We are working on the implementation of Intelligent Wireless ECG Sensor (IWES) by
integrating the cardiac arrhythmia detection algorithm on a chip (ICAC). The ICAC is
currently under evaluation and test on an FPGA. Thus the new platform contains a set of
IWES and a remote server and it will be more reliable and friendly used. With the IWES,
when the patient leaves home (outdoors), the wireless communication is out of range so the
IWES is automatically disconnected to the remote server. Thus, when the cardiac rhythm
disturbances are detected and the ECG signals are recorded locally in the Multimedia flash
card. Therefore only the highest emergency short message will be sent to the remote WAP
server through mobile phone (SMS). The emergency message may be defined by the
cardiologist according to the physical state of the patient. In fact, periodically the IWES
tries to connect to the remote server, if the connection is established a cardiac rhythm
disturbance report or all the recorded events may be sent to the remote server.
5. Acknowledgements
We would like to thank the MENRT, the ANVAR and the Conseil Regional
d’AUVERGNE for their support to this project.
References
[1] P. Palau, K.M. Hou et J. Ponsonnaille, "TelmedTCP : Protocole TCP/IP temps réel dédié à la
télémédecine", revue Informatique et santé, Télémédecine et e-santé, 2002 vol 13, pp 53-60, Springer-
Verlag France.
[2] D. Perrot, M. Dabonneville, K. M. Hou et J. Ponsonnaille, "Plate-forme de validation dédiée à la
télémédecine – Application à la détection d’arythmie dans les signaux ECG", revue Informatique et santé,
Télémédecine et e-santé, 2002 vol 13, pp 149-156, Springer-Verlag France.
[3] C. de Vaulx et K. M. Hou, "DREAM : un micro noyau réparti, temps réel orienté pour la tolérance aux
fautes", revue Informatique et santé, Télémédecine et e-santé, 2002 vol 13, pp 63-69, Springer-Verlag
France.
[4] Zhou H, Palau P, De Vaulx C, Li JJ, Guo C, Quilliot A and Hou KM. A Configurable Micro Network
Kernel dedicated to real-time distributed embedded network sensor: Telemedicine application. 2002 6th
international conference on signal processing proceedings, August 26-30, 2002, Beijing China, pp 1748-
1752.
[5] Jenkins JM and Caswell SA. Detection Algorithms in Implantable Cardioverter Defibrillator. Proceeding
of the IEEE. March 1996; pp. 428-445.
[6] Bailey JJ, Berson AS, Garson A, Horan LG, Macfarlane PW, Mortara DW and Zywietz C.
Recommendations for standardization and Specifications in automated electrocardiography: Bandwidth
and digital signal processing. Special report Circulation, February 1990; 81 (2): pp. 730-739.
[7] Berbari EJ, Scherlag BJ and Lazzara R. A computerized Technique to Record New Components of the
Electrocardiogram. Proceeding of the IEEE. May 1977; pp. 799-802.
[8] Collins SM and Arzbaecher RC. An Efficient Algorithm for Waveform Analysis Using the Correlation
Coefficient. Computers and biomedical research 14, 1981, pp 381-389
[9] Koyrakh LA, Gillberg JM and Wood NM. Wavelet Transform Based Algorithms for EGM Morphology
Discrimination for Implantable ICDs. Computers in Cardiology, 1999, No 26, pp 343-346.
[10] Swerdlow CD et Al. Discrimination of Ventricular Tachycardia from Supraventricular Tachycardia by a
Download Wavelet-Transform Morphology Algorithm: A Paradigm for Development of Implantable
Cardioverter Defibrillator Detection Algorithm. Journal of CARDIOVASCULAR Electrophysiology,
Vol 13, No 5, May 2002.
[11] http://www.novacor.com
[12] Gineste L. Plate-forme de suivi à distance d’arythmie cardiaques. Rapport de DEA CSTI, Université
Blaise Pascal Clermont-Ferrand II, 2002.
[13] Texas-Instruments. MSP430x13x, MSP430x14x MIXED SIGNAL MICROCONTROLLER. July 2000,
pp 1-67.

Page 9

9
[14] http://www.physionet.com
[15] http://www.who.int (Who: World Health group)
[16] K.A. Ellenbogan, Cardiac Pacing, Chapter 1, pp. 1-10, Blackwell Scientific Publications, 1992.
[17] "Sudden death and how to prevent it" A commentary from Internet Medical Education, Inc.
[18] " The Electrocardiogram( ECG)", Richard N. Fogoros, M.D. http://heartdisease.about.com
[19] "Holter Monitors and Event Recorders" Richard N. Fogoros, M.D. http://heartdisease.about.com
[20] C. Partridge, “The End of Simple Traffic Models”, (Editor’s Note), IEEE Network, Vol. 7, No. 5,
September 1993, page3-3.
[21] http://www.intel.com/support/proshare/8150.htm#5


Contact Persons:
Name: ZHOU Haiying
Email: hyzhou@isima.fr
Name: Kun Mean Hou
Email: kun-mean.hou@isima.fr