The 5th International Conference on Electrical Engineering and Informatics 2015

August 10-11, 2015, Bali, Indonesia

Design of ECG Homecare:12-Lead ECG Acquisition

using Single Channel ECG Device Developed on
AD8232 Analog Front End
Muhammad Wildan Gifari1, Hasballah Zakaria2, Richard Mengko3
KK Teknik Biomedika
Sekolah Teknik Elektro dan Informatika ITB
Bandung, Indonesia
1
[email protected], [email protected], [email protected]

Abstract—Electrocardiogram (ECG) devices measure

electrical activity of the heart muscle to determine heart
conditions. ECG signal quality is the key factor in determining
the diseases of the heart. However, the availability of ECG
devices is causing ECG diagnosis slow and difficult. This paper
presents the design of single channel portable ECG device
developed on AD8232 chip platform. To extend the capability of
an ECG portable device, a 12-lead ECG acquistion technique is
also tested out. The results showed the single channel ECG Fig. 1. 12-lead electrode placement, red colour is ground electrode, while
blue and black color is postive and negative electrode, respectively.
module with additional 12-lead ECG acquisition technique can
serve the function of ECG Homecare device, thus making the
Recent research of ECG in Android such as HeartToGo
ECG diagnosis much more accessible.
developed by Zhanpeng Jin [2] is able to detect R peak and
Keywords—portable one-lead ECG device, twelve-lead ECG calculate beat detection. The system also capable of detecting
acquisition technique, ECG Homecare. sinus bradycardia, ventricular flutter, and left bundle branch
block. Mobile ECG developed by Harri Kailanto [3] has
combined the Android with server to enable storage and web
I. INTRODUCTION transmission of ECG data. QRS and arrythmia detection are
Today, heart disease is one of the prominent cause of death the features of Mobile ECG. However, from the two paper the
wolrwide. In Indonesia, roughly 700 thousands mortalities per analysis is limited to only one-lead ECG.
year is caused by heart attack. To prevent heart attack from
becoming severe, early diagnosis is of utmost importance. One In this paper, we design an ECG Homecare device that is
of the diagnostic technique is Electrocardiogram (ECG). ECG capable to acquire the 12-leads ECG signal. The device
devices record electrical signal from cardiac muscle to predict portability is made possible using single channel ECG Analog
the abnormality present in the heart. Front End (AFE) AD8232. Our contribution will be the testing
of twelve-lead ECG acquisition technique that enables single
However, existing ECG devices in Indonesia mostly are channel ECG to help early diagnosis of various diseases.
situated only in big hospitals. The patient with heart attack
will have to wait for the ambulance to come and escort him to II. DEVICE SPECIFICATION
the hospital. While waiting the ambulance, the important
signal associated with the heart attack may be diminishing, A. ECG Signal Characteristics
thus making the doctors loss traces of the causes of the heart
attack. The other patients are too busy to do ECG check-up Before we derive the technical specification, first we
unless something wrong happened to their heart. ECG check- examine the characteristics of ECG signal.
up is bothersome. ECG signal consists of low amplitude signal superimposed
Therefore, it is very important to develop portable ECG on high common voltage and noise [4]. The common voltage
devices. The portability of the devices will make it possible to comes from electrode-skin offset and can be as high as 300
place the devices in community health center (Puskemas) or mV. ECG signal itself has frequency content of 0.05-100 Hz
even in one’s home. The diagnosis will also be much easier with amplitude ranging from 0-3.0 mV. The noises that
and cheaper. The special condition such as patient with heart presents in ECG signals are caused by:
attack will also be accommodated with portable ECG • Powerline interference on 50 or 60 Hz harmonics
Homecare device. Furthermore, it is possible to acquire 12-
lead ECG using single lead ECG [1]. The electrode placement • Electrode offest noise caused by skin-electrode
for 12-lead ECG aquisition is pictured in Fig I. contacts, DC signal of amplitude 300 mV

978-1-4673-7319-7/15/$31.00 ©2015 IEEE

371
 • Muscle contraction caused by patient movement starts with level 0 design, continued by level 1 design, and so
on until every level can be implemented using specific
• Baseline wander caused by patient respiration, hardware.
low frequency AC signal
• Electromagnetic interference from another A. Design Overview
electronic devices, usually high frequency signal

B. Technical Specification
According to Rachit [5], ECG bandwidth specification is
dependent with its application. The bandwidth specification is
presented in table 1.
Fig. 2. Level 0 Functional Block of ECG Homecare Device
TABLE I. ECG APPLICATION AND ITS FREQUENCY BANDWIDTH

Application Bandwidth (Hz) TABLE III. ECG HOMECARE BLOCK DESCRIPTION

Display 0.5-40
Module ECG Homecare Device
QRS Detection 0.5-40 Input Heart Voltage (0.01-250 Hz, 0-3 mV)
Arryhtmia detection 0.05-60 Output Serial Data
ST segment monitoring 0.05-60 Conditioning heart voltage (filtering,
Function amplifying, digitizing) to be serial data
which is sent to PC or Smartphone
In this paper, we develop Homecare ECG device, which is
used primarily for display and QRS detection. Hence,
bandwidth of 0.5-40 Hz is chosen. The four electrodes system The level 0 block diagram in Fig. 2 is analyzed further and
is employed to be able to acquire the augmented leads. Input is defragmented in level 1 block diagram in Fig. 3.
provided by 9-V home battery. For complete specifications
please refer to Table II. ECG signal from patient is first tapped using three
electrodes for all leads except augmented leads, where we use
one additional electrode. Bandpass filter and amplifier are
TABLE II. ECG HOMECARE SPECIFICATION implemented using integrated Analog Front End (AFE) ECG
Parameter Value chip. Microcontroller handles the ADC conversion, and
bluetooth transmitter/serial cable transmit digital data to
Application QRS Detection & Display smartphone/PC. All system is powered by a 9V battery.
Channel Single channel
B. Implementation
Lead Reading Up to 12 leads w/ 4 electrodes
• Electrode is implemented using gel electrode.
Bandwidth 0.5-40 Hz Eventhough gel electrode is disposable, it is cheaper
CMRR 80 dB and more convenient to use, because no external gel is
needed. Another electrode, clamp electrode is more
Resolution Better than 5 μV expensive and needs extra gel for good conductivity.
Gain More than 1000 V/V
• HC-05 is used for Bluetooth transmission. HC-05 uses
Sampling Frequnecy Faster than 250 samples / second Bluetooth version 2.0 Its data speed range from 9600-
460800 bps. We choose 10-bit ATMega in Arduino
Input Voltage 5-12 V
Uno R3 board for the microcontroller. The
Connectivity Serial Cable and Bluetooth microcontroller is powered by 9V battery. Connectivity
to PC is achieved using Arduino serial cable.
III. SYSTEM DESIGN AND IMPLEMENTATION
• Bandpass Filter and Amplifier is explained in section
To tackle all the specification as mentioned in Table II, we C.
design the hardware using step-by step method. The design

Fig. 3. Level 1 Functional Block of ECG Homecare Device

372
 TABLE VII. AFE ECG CHIP COMPARISON
C. Bandpass Filter and Amplifier
Chip
Parameter
AD8232 HM301D ADS1191
ST Texas
Company Analog Devices
Microelectronics Instruments
CMRR 80 dB 100 dB 95 dB
Fig. 4. Level 0 Functional Block of Bandpass Filter + Amplifier Output
10 Gȍ 50 Mȍ 100 Mȍ
Impedance
Gain 100 V/V 64 V/V 12 V/V
TABLE IV. BANDPASS FILTER + AMPLIFIER BLOCK DESCRIPTION Low noise
Rail to rail
Feature 3-channel ECG PGA & high-
Module ECG Homecare Device output
res ADC
Input Heart Voltage (0.01-250 Hz, 0-3 mV) Price $ 19.95 (board)
$ 125 (evaluation
$ 7.96 (chip)
board)
Output ECG Signal 0-3.3 V, 0.5-40 Hz
¾ Filtering high frequency and low
AD8232 is preferred over another chips. HM301D is three
frequency component channel, while we only need single channel ECG. ADS1191
Function
¾ Provide >1000 V/V voltage gain doesn’t provide high enough gain to get good resolution.
¾ Rejecting common mode signal AD8232 has the best output impedance and gain. The high-
pass and lowpass filter designs are available in AD8232
One of the solution for realizing block in Fig. 4 is using datasheet [6]. For highpass filter, two-pole HPF is used and
cascade of instrumentation amplifier and operational for lowpass filter, two-pole Sallen-Key LPF is used.
amplifier. Instrumentation amplifier has DC rejection and SparkfunTM provides readily available 0.5-40 Hz ECG board
differential amplification features, which makes with AD8232. We implement the Active Filters using this
instrumentation amplifier serves the function of highpass integrated board.
active filter. For the second stage, operational amplifier is
implemented to achieve lowpass active filter. These design D. Shield
solution is pictured in Fig. 5. All the hardware above need to be packed in compact
form. A shield is desgined to place all the component together.
The shield is stacked right above the Arduino. Battery,
Bluetooth, and ECG chip are stacked above the shield. Battery
supplies the power to Arduino board while the Arduino 3.3 V
output supplies the power to HC-05 and AD8232 (see Fig. 6).

Fig. 5. Level 1 Functional Block of Bandpass Filter + Amplifier

TABLE V. HIGH PASS ACTIVE FILTER BLOCK DESCRIPTION

Module ECG Homecare Device

Input Heart Voltage (0.01-250 Hz, 0-3 mV)
Output ECG Signal 0.5-250 Hz, amplified
¾ DC Blocking
Function
¾ Provide first stage amplification

TABLE VI. LOW PASS ACTIVE FILTER BLOCK DESCRIPTION

Fig. 6. Left: design of PCB Shield, right: ECG Homecare prototype
Module ECG Homecare Device
Input Heart Voltage (0.05-250 Hz, amplified) E. Software
Output ECG Signal 0.5-40 Hz, 0-3.3 V • Arduino IDE is used to program the microcontroller
¾ Low Pass Filtering Becaues it has easier languange to implement. The
Function program in microcontroller is used to detect lead and
¾ Provide second stage amplification
send the data.

There are a number of available analog front end (AFE) • Processing is software on PC. Processing code is used
ECG chips available in the market. Comparison of several to graph the incoming data and store the data in .txt
chip can be seen in Table VII [6,7,8]. format along with its timestamp

373
 • Bluetooth Graphics is an apps on Android. Bluetooth Lead Subject One Subject Two
graphics display the real-time data and also has
abs (0.1117 ± 0.10) (0.0338 ± 0.05)
capability to log the incoming data. I
rel (18.62 ± 17)% (7.521 ± 11)%

IV. TESTING abs (0.5001 ± 0.10) (0.4592 ± 0.05)

II
rel (20.84 ± 4.2)% (43.74 ± 4.8)%
A. Wireless and Wired Data Transmission
abs (0.5978 ± 0.10) (0.5062 ± 0.05)
Wireless data is sent via Bluetooth. AsusTM Zenfone 5 III
graphs and logs the data using Bluetooth GraphicsTM apps. For rel (24.91 ± 4.2)% (67.50 ± 6.7)%
wireless data transmission, clean signal is difficult to obtain. abs (0.1459 ± 0.10) (0.3076 ± 0.05)
Wired data transmission results in much clearer signal (see aVr
rel (11.22 ± 7.7)% (43.95 ± 7.1)%
Fig. 7). In wired data transmission, the drawback is the data
can become unstable. Faster data corrresponds to shorter abs (0.1471 ± 0.10) (0.2859 ± 0.05)
aVl
recording time. rel (11.31 ± 7.7)% (95.31 ± 17)%
abs (0.6800± 0.10) (0.5127 ± 0.05)
aVf
rel (27.20 ± 4.0)% (60.32 ± 5.9)%
abs (0.4466 ± 0.10) (0.7855 ± 0.10)
V1
rel (31.90 ± 7.1)% (41.34 ± 5.3)%
abs (0.7179 ± 0.10) (0.4651 ± 0.10)
V2
rel (27.61 ± 3.8)% (16.61 ± 3.6)%
(a) abs (0.4772 ± 0.10) (0.2367 ± 0.10)
V3
rel (26.51 ± 5.6)% (8.160 ± 3.4)%
abs (0.4772 ± 0.10) (0.4145 ± 0.10)
V4
rel (26.51 ± 5.6)% (18.84 ± 4.55)%
abs (0.9328 ± 0.10) (0.2090 ± 0.10)
V5
rel (40.56 ± 4.3)% (9.953 ± 4.8)%
(b) abs (1.414 ± 0.10) (0.1586 ± 0.10)
V6
Fig. 7. ECG signal display (a) in Android, (b) in PC rel (64.29 ± 4.5)% (15.86 ± 10)%
Abs = absolute error (mV), rel = relative error (%)
TABLE VIII. WIRED DATA TRANSMISSION PERFORMANCE
V. DISCUSSION
Baud rate Data rate
Notes Purpose of the design, which is to acquire 12-lead ECG
(bps) (sample / second)
9600 192 Very stable using single channel ECG is fulfilled. Prototype has been
made and tested against standard clinical ECG. For most lead
Data unstable after one on the two subjects, the waveform acquired has similar shape
14400 288
minute with the standard device (see appendix). For amplitude
Data unstable after 42 accuracy, the best that can be achieved is 7 % and the worst is
second 95 %, with average accuracy 33%. The cable transmission
19200 365
Data rate varies from results in clean signal. However, the Bluetooth data
331-386 sps transmission need to be debugged to obtain cleaner signal.
B. Comparison with Clinical ECG The ECG signal obtained by ECG Homecare device are
The data is transmitted at 19200 bps baud rate. Clinical stored as .txt file which is compliant with MIT-BIH standard.
ECG used to compare the data is Nihon KohdenTM Cardiofax The signal can be processed by Java application, which my
9620. Nihon Kohden has sample rate 500 sps and bandwidth two laboratory mates are currently making. Processing the 12-
0.05-75 Hz, with notch filter on 50 Hz [9]. Data acquistion is lead ECG enables the early detection of various heart
carried out while patient is on sleeping position, minimizing abnormalities.
motion artifacts. Laptop is not connected to powerline to Futher development should concern with the development
prevent AC 50 Hz noise. For each subject, 12-lead ECG is of low power ECG device that can last long enough for a day.
taken. Table IX showed the quantitative comparison between Accuracy needs to be improved. Adjustment of filter
data acquired with ECG homecare device and standard device. bandwidth can be implemented to provide multi purpose
portable ECG. Internal memory is useful to store the ECG data
TABLE IX. QUANTITATIVE COMPARISON BETWEEN DATA when the user doesn’t bring along any laptop or Android.
ACQUIRED WITH ECG HOMECARE AND NIHON KOHDEN

374
 REFERENCES Le
ECG Homecare Nihon Kohden
ad
[1] http://www.amperordirect.com/pc//12lead-home-ekg-monitor.html.
aVf
Accessed on March 24, 2015
[2] Z. Jing, J. Oresko, S. Huang, A.C. Cheng, “Heart to Go: A Personalized
Medicine Technology for Cardiovascular Disease Prevention and
Detection,” IEEE Life Sciences System and Application Workshop
(LiSSA), 2009
[3] H. Kailanto, E. Hyvarinen, J. Hyttinen, “Mobile ECG Device
Measurement and Analysis System Using Mobile Phone as the Base
Station,” Proceedings of the 2nd International Conference of Pervasive
Computing Technologies for Healthcare, 2008
[4] Ajay Bharadwaj & Umanath Kamath, “Accurate ECG Signal V1
Procesing,” CA: Cypress Semiconductor, 2011.
[5] Rachit Mochan, “Fully Integrated Analog front-end for a 2 electrode
ECG device,” M.Sc Thesis, Delft: Technical University of Delft, 2011.
[6] AD8232 Datasheet, Analog DevicesTM
[7] ADS1191 Datasheet, Texas InstrumentTM
[8] HM301D Datasheet, ST MicroelectronicsTM
[9] Nihon Kohden Cardiofax 9620 Repair Manual, Japan: Nihon KohdenTM
V2
ATTACHMENT

SAMPLE ECG DATA ACQUISTION FOR SUBJECT ONE

Le
ECG Homecare Nihon Kohden
ad
I
V3

II

V4

III

V5

aVr

V6

aVl

375
 SAMPLE ECG DATA ACQUISTION FOR SUBJECT TWO Le
ECG Homecare Nihon Kohden
ad
Le
ECG Homecare Nihon Kohden V1
ad
I

II V2

V3
III

V4
aVr

aVl
V5

aVf

V6

376