Inexpensive, Portable, Smartphone-

Based 12-Lead Electrocardiogram

By:

________________________________________
Jordon Friedman

________________________________________
Sean Murphy

________________________________________
Blaine Rieger

Date: October 14, 2015

Major Qualifying Project Proposal submitted to the Faculty of

WORCESTER POLYTECHNIC INSTITUTE

In partial fulfillment of the requirements for the

Degree in Bachelor of Science

Date: 18 December, 2015

Approved:

____________________________
Professor Reinhold Ludwig, Advisor

____________________________
Professor John McNeill, Co-Advisor

1
Contents
Table of Figures ........................................................................................................................................ 3
Abstract .................................................................................................................................................... 4
Chapter 1 – Introduction .......................................................................................................................... 4
Chapter 2 – Background ........................................................................................................................... 5
2.1 – Definition and History of the 12 ECG Leads ................................................................................. 5
2.2 – ECG Physiology ............................................................................................................................ 9
2.3 – Current ECG Technology ............................................................................................................ 12
Chapter 3 - Project Objectives ................................................................................................................ 12
High-Level Architecture ...................................................................................................................... 12
ECG Analog Front End Board .............................................................................................................. 13
Embedded Board and Bluetooth Connectivity ................................................................................... 13
Android Application............................................................................................................................ 14
Chapter 4 - Approach ............................................................................................................................. 14
Step 1 – Background Electronics Research ......................................................................................... 14
Step 2 – Expansion and Verification of the Analog Front End (AFE) .................................................. 15
Step 4 – Embedded Software ............................................................................................................. 20
Step 5 – Android Development .......................................................................................................... 21
Step 6 – Functionality Test ................................................................................................................. 22
Chapter 5 - Project Deliverables ............................................................................................................. 23
Chapter 7 - Conclusion ....................................................................................................................... 23
Bibliography ............................................................................................................................................... 25

2
Table of Figures
Figure 1 - ECG BLOCK DIAGRAM ................................................................................................................... 5
Figure 2 – 10-leads of waller (A) and einthoven limb leads (B) .................................................................... 6
Figure 3 - WILSON CENTRAL TERMINAL ....................................................................................................... 7
Figure 4 - GOLDBERGER AUGMENTED LEADS (a), THEIR LOCATION (B), AND PRECORDIAL LEAD
PLACEMENT .................................................................................................................................................. 8
Figure 5 - PROJECTION SO THE 12 LEAD VECTORS ....................................................................................... 9
Figure 6 - NORMAL SINUS RHYTHM ON AN ECG ........................................................................................ 10
Figure 7 - ELECTRICAL CONDUCTION SYSTEM OF THE HEART ................................................................... 11
Figure 8 - HIGH LEVEL ARCHITECTURE ....................................................................................................... 13
Figure 9 - TYPICAL OUTPUT OF A 12-LEAD ECG .......................................................................................... 14
Figure 10 - SINGLE-SUPPLY, VERY LOW POWER, ECG CIRCUIT ................................................................... 15
Figure 11 – INPUT BUFFERS AND WILSON CENTRAL TERMINAL ................................................................ 17
Figure 12 - INPUT BUFFER FREQUENCY RESPONSE .................................................................................... 17
Figure 13 – OUTPUT STAGE WITH ACTIVE FILTERING ................................................................................ 18
Figure 14 - SIMULATED OUTPUT WAVEFORM @100Hz ............................................................................. 18
Figure 15 - OUTPUT STAGE FREQUENCY RESPONSE .................................................................................. 19
Figure 16 – RIGHT-LEG DRIVE CIRCUIT (RLD) ............................................................................................. 19
Figure 17 - RIGHT LEG DRIVE FREQUENCY RESPONSE................................................................................ 20
Figure 18 - EMBEDDED SOFTWARE DIAGRAM ........................................................................................... 20
Figure 19 - ANDROID SOFTWARE FLOWCHART .......................................................................................... 21
Figure 20 - SAMPLE 12-LEAD ECG OUTPUT ................................................................................................ 22

3
Abstract
The electronics industry is constantly focusing on creating cheaper, more portable electronics that
integrate with smartphone technology and the increasing demand for affordable and effective medical
instrumentation. Furthermore, the healthcare industry is beginning to follow suit, with many medical
devices moving to mobile-based platforms and hardware. This Major Qualifying Project aims to
complete preliminary development of an inexpensive, accurate, portable, Bluetooth 12-lead
electrocardiogram (ECG) that interfaces with any Android smartphone. The current market lacks in
mobile ECGs for portable devices to use clinically and other portable ECGs are extremely expensive. This
project aims to use a compact analog front-end board, an embedded processor with Bluetooth adapter,
and generic Android smartphone to revolutionize the way the healthcare industry envisions portable 12-
lead ECGs.

Chapter 1 – Introduction
We live at an exciting time in the technology industry. Despite the fact that modern technological
devices such as phones, computers, and various peripherals are getting cheaper and smaller, they
continue to become more powerful, more accurate, and more portable. Combining these observations
with the fact that science is continually discovering new methods for acquiring and processing data, we
find that individuals are using today’s smartphone technology to create innovative solutions to issues
that would not have been solvable otherwise.

One of the most prominent issues in medical technology today is the cost of medical devices and, more
specifically, how much individuals are able to spend on average to purchase life-saving diagnostic
equipment. According to the World Health Organization (WHO), developed countries such as the United
States spend $290 per capita on medical equipment, whereas developing countries such as India only
spend a miniscule $6 per capita on medical equipment.1 Devices such as electrocardiograms (ECG/EKG)
are used to measure the electrical activity of the heart and diagnose life-threatening conditions such as
myocardial infarctions (MI), or heart attacks. The cost of such devices currently range from about $2,000
and $8,000 for simple units to over $13,000 for portable units that contain defibrillators.2 3 Given these
huge costs, it is apparent that there needs to be more affordable alternatives created for the healthcare
markets in countries that simply cannot spend that much on medical devices.

Given the current environment of both consumer electronics and medical devices, it was inevitable that
an increasing number of corporations and startups would work toward developing low-cost,
smartphone-based medical devices. From Bluetooth blood pressure cuffs4 to watches that measure your
blood-oxygen saturation5 to glucometers that report your blood glucose to your phone6, there are new

1
(Nimunkar, David Van Sickle, Van Sickle, & Webster, 2009)
2
(Johnson, 2011)
3
(DXE Medical, 2015)
4
(Withings, 2015)
5
(Withings, 2015)
6
(iHealth, 2015)

4
devices practically every day that try to blend the realm of consumer electronics and medicine to
improve modern healthcare both in and out of the clinical setting.

FIGURE 1 - ECG BLOCK DIAGRAM7

In this project, we aimed to combine our skills in electrical engineering and software engineering with
our clinical knowledge of the human cardiovascular system to develop a low-cost, smartphone-based
ECG. Figure 1 outlines the basic block diagram of an ECG. The ECG is capable of pairing to any Android
smartphone using Bluetooth to allow physicians and other healthcare providers in practically any setting
at practically any socio-economic position to gather extremely important clinical information and
transmit it anywhere in the world over the internet. This will allow healthcare providers tools to better
diagnose the chief complaints of patients and achieve an improved differential diagnosis using essential
data that they most likely would not have had access to previously.

Chapter 2 – Background
2.1 – Definition and History of the 12 ECG Leads
The ECG is hardly a new piece of technology. Augustus Désiré Waller, a British physiologist from the late
19th century, experimented with bio-electromagnetism, which lead him to develop the first
electrocardiogram in 1887. Unlike modern ECGs, which use the right arm, left arm, and left leg to get
signals, Waller used these three positions, as well as the right leg and mouth, to get ECG signals. An ECG
ultimately displays what are referred to as leads. Leads are differential signals, which compare the
difference in potential (voltage) between different points on the body. Waller’s system ultimately
created 10 leads between each of his 10 body positions.8

It was not until 1908 when Willem Einthoven, a Dutch doctor and physiologist, published a paper
describing the electrode positions that we use today to acquire data for the limb leads on an ECG. These

7
(Chen, et al., 2013)
8
(Malmivuo & Plonsey, 1995)

5
three leads are appropriately named the Einthoven Lead System. The graphic below outlines Waller’s
lead system (A) as well as the modern Einthoven (B).9

FIGURE 2 – 10-LEADS OF WALLER (A) AND EINTHOVEN LIMB LEADS (B)10

As can be seen in the above graphic, Einthoven’s electrode placement defines the first three of our
twelve-lead ECG system. Those first three leads, referred to as the limb leads, are Lead I, Lead II, and
Lead III. The limb leads are defined as follows11:

Lead I: VI = Φ L - Φ R (1)

Lead II: VII = ΦF - ΦR (2)

Lead III: VIII = ΦF - ΦL (3)

In these equations, 𝑉3 , where n is the number of each respective lead, represents the differential
voltage of lead n and 𝜙3 represents the potential (voltage) at the electrode location on the skin. We can
therefore deduce that, according to Kirchhoff's Voltage Law, 𝑉5 +𝑉555 = 𝑉55 . Therefore, this means that
only two of the three leads are independent. To form the remaining nine leads of the twelve-lead ECG,
we have to use the average of the three limb electrodes to create the Wilson Central Terminal (CT).

Frank Norman Wilson, a professor of internal medicine at the University of Michigan, investigated how
to define unipolar potentials on the ECG, since the limb leads are all bipolar. Although his work did not

9
ibid
10
ibid
11
ibid

6
allow him to gather any information from any purely unipolar potentials, he ultimately developed the
Wilson Central Terminal as a way to create pseudo-unipolar leads. In order to do this, three equal-size
resistors are placed between each of the three limb electrodes and the central junction. As is shown in
Figure 3, this terminal averages to a potential that, when considering the body as a plane, creates a
triangulated point over the heart. This resistor network is outlined in Figure 3a where arbitrary resistor
values shown.

FIGURE 3 - WILSON CENTRAL TERMINAL12

Using this central point of potential, we are able to acquire both the three so-called Goldberger
Augmented Leads and the six Precordial Leads. With these remaining nine leads, we have all twelve of
the leads for our ECG. The Goldberger Augmented Leads are named in honor of Dr. Emanuel
Goldberger, a Manhattan cardiologist and researcher, who discovered in 1942 that, by omitting one of
the three resistors in the Wilson Central Terminal, one can create three additional leads which, due to
the nature of their orientation, are considered unipolar augmented leads.13 These additional three leads
use the existing three limb leads to gain three additional perspectives on the heart’s electrical activity,
as noted in the first image below. In addition to creating the Wilson Central Terminal, Wilson added the
six Precordial Leads in 1944, which added the final six perspectives provided by the modern twelve-lead
ECG. These leads, named V1, V2, V3, V4, V5, and V6, are simply defined as the voltage differential between
the Wilson Central Terminal and the respective Precordial Lead. These six leads, which are placed in
order starting at both sides of the sternum and continuing to a position just between the mid-clavicular

12
ibid
13
(Emanuel Goldberger, Cardiologist, Dies at 81, 1993)

7
and mid-axillary lines, allow the ECG to see the electrical activity on the anterior and left-inferior sides of
the heart.

FIGURE 4 - GOLDBERGER AUGMENTED LEADS (A), THEIR LOCATION (B), AND PRECORDIAL LEAD
PLACEMENT14

The twelve leads of the ECG ultimately grant the user the ability to obtain a three-dimensional
electrostatic representation of the heart, which is illustrated in Figure 5. The following graphic helps to
describe the perspective offered by all twelve leads. If we were to imagine a plane that was created
along the vector of each lead and then were to follow an electrical impulse from the heart’s Sinoatrial
Node to the Atrioventricular Node through the Bundle of His to the Purkinje Fibers, the ECG would
create a trace following this electrical impulse along that plane.

14
(Malmivuo & Plonsey, 1995)

8
 FIGURE 5 - PROJECTION SO THE 12 LEAD VECTORS15

2.2 – ECG Physiology

As was described at the end of the previous section, the ECG follows the electrical activity in the heart.
Figure 6 outlines what we observe on an ECG for a normal sinus rhythm, the normal electrical activity in
a properly functioning heart. The normal sinus rhythm is broken up into a number of waves, the most
prominent are the P, Q, R, S, and T waves. U waves are typically not seen on the ECG. P waves signify
atrial depolarization, the QRS complex signifies ventricular depolarization, and the T waves signify
ventricular repolarization. Atrial depolarization is hidden behind the QRS complex. As can be seen in the
graphic below, in order to get all of the necessary detail for an ECG, the analog system has to be able to
detect voltage changes as small as approximately 0.1mV in the Q wave. Each wave on the ECG
corresponds to a crucial phase of a heartbeat.

15
ibid

9
 FIGURE 6 - NORMAL SINUS RHYTHM ON AN ECG16

When a heart beats, it is broken into a number of specific actions and phases. The action of the
myocardium (cardiac muscle) contraction, and subsequently pumping blood out of the contracting
chamber, is referred to as depolarization. The P wave is the result of the depolarization of the heart’s
atria. The Q, R, and S waves together are referred to as the QRS complex. The QRS complex signifies the
depolarization of the heart’s ventricles. The repolarization (or return to rest) of the heart’s atria is
hidden behind the electrical activity from the QRS complex. The T wave is the repolarization of the
ventricles. As can be seen in the image above, there are a number of different time segments, which can
be used by medical professionals to diagnose different conditions of the heart.

In order to generate these potentials through the heart, which lead to the heart’s depolarization and
subsequent repolarization, the heart has an electrical system that is shown in detail in Figure 7. The
pacemaker of the heart, the Sinoatrial Node (SA node), creates the first electrical stimulus of the heart.
This impulse travels through the atria, causing atrial depolarization, ultimately gathering again at the
Atrioventricular Node (AV node). From here, the impulse gets sent down the Bundle of His and down the
left and right bundle branches. At the bottom of the branches, the impulse is sent to the left and right
ventricles. As the impulse passes through the conductive pathways, it causes the depolarization of the
ventricles. Meanwhile, as the ventricles are contracting, the atria are repolarizing. After the ventricles
contract, they slowly repolarize back to the state of rest.

16
ibid

10
 FIGURE 7 - ELECTRICAL CONDUCTION SYSTEM OF THE HEART17
Although the actions of the myocardium tissue described above are what we see in an ECG, there is a
significantly greater amount of action occurring at the cellular level that leads to the ECG output. As the
electrical impulse from the SA node travels through the heart, it is stimulating the myocardial cells.
What actually happens during this stimulation is the activation of a sodium-potassium pump in the cells.
As the myocardial cells conduct and then pass the signal along to adjacent cells, electrically stimulated
sodium channels are being opened up. These fast sodium channels take in positive sodium ions (Na+),
which surround the cells at rest. Eventually, this rapid influx of positive sodium ions causes the
depolarization of the cell to reach a threshold, causing the cellular action and contraction. The cell
achieves depolarization because the potassium channels, which expel positive potassium ions (K+) from
the cell, are much slower than the sodium channels. This causes the depolarization to be drastically
positive until the potassium ions are expelled. This rapid and large change in cellular charge is what
creates the change in electrical potential on the ECG output.

17
(University of Rochester Medical Center, 2015)

11
2.3 – Current ECG Technology
ECG technology, although all designed to accomplish the same objective, varies greatly between the
hospital environment, the pre-hospital environment, and the home environment. In the hospital
environment, there are large, expensive, powerful devices that can be used to get resting
measurements and others that are designed for actives measurements, such as those done during a
stress test. One example of these modern resting ECG machines is the Philips PageWriter TC70 16-lead
ECG. This is a relatively expensive device with an MSRP of $9990, which incorporates a large cart and
battery for hospital use. These ECGs allow for additional precordial leads for hospital use so as to
diagnose right-side myocardial infarction involvement. Furthermore, they contain very powerful
processors for advanced DSP that simply is not practical for use in the smaller, less expensive devices.
However, these devices allow for database integration over Wi-Fi so that clinicians can easily access old
ECGs. Although typically only 12-lead ECG, the active ECGs for stress tests typically cost more than the
resting ECGs.18

In the pre-hospital environment, devices tend to get significantly more expensive. Although these
devices tend to incorporate a number of additional features such as blood-oxygen saturation, carbon-
dioxide concentration, automatic blood pressure monitors, biphasic defibrillators, and pacers, they
typically lack the resolution of the hospital-grade equipment. These portable units are built extremely
robust to take falls and physical damage in the pre-hospital environment. Despite their lack of accuracy
compared to the hospital equipment, they are still accurate enough to provide life-saving differential
diagnoses on patients in the field. These portable units typically cost between $10000 and $50000.19

In the wireless ECG environment, there are very few options that are available at the moment. There are
a number of products currently available that are able to acquire a single-lead ECG and wireless transmit
to a device to view, but there are few to none that are able to interface with a smartphone over
Bluetooth. The only current example of a company trying to achieve a smartphone-based 12-lead ECG is
Nimbleheart Inc. in California. However, none of their proposed products are currently market-ready.
No pricing is presently available to serve as a reference. Despite being in this early state, the company is
pursuing a FDA Class II certification for its device. The cost of this certification is expected to be
extremely high and will likely exclude the medical device markets in developing countries due to
exorbitant cost like most other devices currently available.20

Chapter 3 - Project Objectives

High-Level Architecture
In this project, we aim to create a novel architecture for an Android-based 12-lead ECG that is capable of
streaming live ECG data to a smartphone with an accuracy that is comparable to that of a modern

18
(Philips, 2015)
19
(Physio-Control Inc., 2015)
20
(Nimbleheart, 2015)

12
hospital-grade ECG. In order to do this, there are three major components to the overall architecture
which are outlined in (INSERT) with their respective interfaces. The first and arguably most important is
the ECG analog front end (AFE). This AFE will be responsible for acquiring all ECG signals in the analog
realm in real time with tremendous accuracy. The second major part of the architecture is the
embedded processor board. This board will be able to read in the 12 analog outputs from the AFE via
analog-to-digital conversion at a rate that best preserves the resolution of the acquired lead signals. This
board will then have to interface with a smartphone over Classic Bluetooth at a rate fast enough to keep
the acquired signals real-time. Finally, a smartphone running the Android operating system will read in
the data received through Bluetooth and display this data in real time on the phone’s display.

FIGURE 8 - HIGH LEVEL ARCHITECTURE

ECG Analog Front End Board

The analog front end of the device is the most crucial part of the system as this is responsible for the
initial signal acquisition for the leads. As was discussed in a prior section, the signals that have to be
acquired range from approximately 0.1mV to approximately 1mV. These are extremely small values so
we will need to use medical-grade components. Typically speaking, most ECG systems use a total system
gain of 1kV/V for each of the output leads. In order to achieve high accuracy, high gain, and high
resolution, we will need to have to use the most accurate resistors for gain while using operational
amplifiers with an input offset voltage on the order of picovolts. Texas Instruments makes a number of
components that are specifically designed for this type of signal acquisition. Finally, another important
aspect of the AFE board is the right-leg drive circuit. A RL-drive circuit is a feature commonly used in
electrical medical instruments such as EEGs and ECGs to reduce noise and to set the common-mode
ratio voltage of the body. This is a feedback loop that takes the average voltage from the acquired
signals, offsets them to the median voltage of any ADC’s voltage (2.5V in this case), and amplifies them
at -38 to reduce the body’s noise. We aim to keep the cost of this under $300 for a single board, not
taking economies of scale into account. We will simply use an off-the-shelf set of electrodes to plug into
the board.

Embedded Board and Bluetooth Connectivity

For our embedded board, we aim to read the 12 analog inputs through 12 unique analog-to-digital
converters on a Cortex M4 processor into buffers with a resolution of at least 12-bit accuracy at a speed
of at least 1MS/s. in order to give us the greatest level of flexibility since the AHA only requires us to
sample at 500S/s for an ECG. These buffered inputs are transmitted into time-divided blocks over Classic
Bluetooth using a Bluetooth capable expansion board to an Android smartphone. The smartphone will
be running a bespoke application that is designed to read our data stream. This board, along with the
AFE board, will run off a 5V lithium-ion rechargeable battery pack that can be charged over USB. This
entire unit will be housed in a 3D printed enclosure. If money and time permit, we would like to create

13
an all-inclusive printed circuit board (PCB) that contains the ARM processor, AFE for lead acquisition,
and any wireless hardware.

Android Application
The Android smartphone application, as previously stated, will be the end-point for data acquisition. The
app will be designed to take in a Bluetooth serial data stream, determine which lead it belongs to, and
append it to the current display of real-time ECG data. The app will also be responsible for taking
standard ECG time measurements such as ventricular rate, PR interval time, QRS duration, among
others. The app will also be capable of “printing” out the ECG acquisition. It will look similar to the
standard ECG output shown in Figure 9. This image will have the ability to be emailed to other
physicians. There may have to be some research done in how to make this output HIPAA compliant.
Given extra time, we may add in rhythm detection to detect rhythms such as atrial fibrillation, ST-
Elevation Myocardial Infarctions (STEMIs), heart blocks, and more.

FIGURE 9 - TYPICAL OUTPUT OF A 12-LEAD ECG21

Chapter 4 - Approach
Step 1 – Background Electronics Research
Our design approach is straightforward. When creating our ECG circuit, we are relying on existing
circuitry. Specifically, the circuit to acquire 12 ECG leads is widely published and a number of different
component manufactures either make application specific IC’s that output SPI for this purpose or make
instrumentation amplifiers and operational amplifiers that are intended for use in ECG applications. As a
design challenge, we are going to make the entire analog front end. As an initial reference, we based

21
(Larson, 2014)

14
our 12-lead design on a single-lead, single-supply, very low power ECG circuit from the OPA333 data
sheet, shown in Figure 10.22

FIGURE 10 - SINGLE-SUPPLY, VERY LOW POWER, ECG CIRCUIT23

Step 2 – Expansion and Verification of the Analog Front End (AFE)

As can be seen in Figure 10, TI has generated a circuit for a single-lead ECG acquisition like those found
in hospital bedside monitors. A single-lead ECG uses three sensing leads (RA, LA, LL) and compared two
to create its single output lead. It also uses a right-leg drive circuit (RL) to set the common-mode voltage
and reduce noise in the body. Using the knowledge we gained in pre-project research that can be found
in our background section, we were able to appropriately expand this circuit to include all 12 ECG
signals. The output stage is the same for each of the 12 output leads as this allows each output lead to
have the same frequency response and. The difference lies mainly in the use of the correct input for the
INA321 instrumentation amplifier. To verify this circuit, we use the analog analysis software Multisim to

22
(Texas Instruments, 2013)
23
(Texas Instruments, 2013)

15
confirm that the output of the circuit functions as anticipated and within the correct frequency response
required for an accurate ECG (frequencies below 150Hz), as defined by the AHA.24

In order to achieve expected functionality from the analog front-end board outlined in Figure 10 -
SINGLE-SUPPLY, VERY LOW POWER, ECG CIRCUIT through Figure 15, there needed to be some relatively
simple hand analysis. The input buffers simply function such that Vin = Vout. The Wilson Central Terminal
is a voltage averaging circuit of the limb-lead signals. Vout is defined as follows:
9: == 9:
< <
9; 9> 9?
V678 = ; ; ; (4)
< <
9; 9> 9?

The resistor values used in the averaging circuit are arbitrary. However, evenly averaging all three
signals against one another requires that all three resistors be equal. For the ECG, we are using resistors
of 100kΩ. As was defined above, the AHA recommends approximate frequency bandwidth of <150Hz to
achieve correct ECG functionality. We achieve this using an active low-pass filter on the output amplifier
stage of the ECG, which is shown in Figure 13. This is simply an RC circuit for which the analysis is shown
as follows:
B B
fA = = = 159Hz (5)
CDEF CD BGΩ BIJ

As is shown here, the active low-pass filter used on the output will provide a cut-off frequency of 159Hz
when using the resistor value of 1MΩ and the capacitor value of 1nF. This allows the device to meet the
AHA guideline on frequency bandwidth. Most right-leg drive circuits also contain an active low-pass
filter so that unnecessarily high frequencies are not driven back to the body when attempting to reduce
noise. Our right-leg drive circuit’s filter is defined as follows:
B B
fA = = = 8.682kHz (6)
CDEF CD OPQRΩ STUJ

This cutoff frequency allows reasonable noise reduction with filtering for abnormally high frequencies
that do not need to be driven to the body. The final analysis determined what resistors to use to get the
desired output gain of the output stage shown in Figure 13. It is important to note that the INA2321 has
a default gain of 5, which means that the output gain that is set is multiplied by 5. In order to get desired
ECG output, we needed a gain of 1000, or 1kV/V. The simulated output stage showing the simulated
gain of 1000 is shown in Figure 14. This is shown as follows:
\]^_ EC BGΩ
A[ = =5∗ =5∗ = 1000𝑉/𝑉 (7)
\`a EB cRΩ

Figure 15 shows that we have the expected active low-pass filter response with the knee positioned at
150Hz. Figure 11, Figure 13, and Figure 16 show our solution to generating inputs, creating the Wilson
Central Terminal, generating outputs, and using the Right-Leg Drive circuit to reduce system noise and
set the Common-Mode Ratio Voltage of the body to 0.5*Vcc. The input stage in Figure 9 involves using
buffers to separate the High-Z impedance of the human body from the rest of the circuit. It also shows
how the limb leads are combined, as previously defined, to create the Wilson Central Terminal. The
output and amplification stage in Figure 10 shows us using an INA321 instrumentation amplifier and two
OPA333 operational amplifiers to get our output signal, do active filtering on the output to meet our

24
(Rijnbeek, Kors, & Witsenburg, 2001)

16
frequency response requirement, amplify the output with a gain of 1kV/V, and set the reference voltage
to that of the body (currently 0.5*Vcc due to the RLD circuit). Finally, we can see how our right-leg drive
circuit in Figure 11 uses two OPA333 operational amplifiers to buffer the output, invert the signal from
the WCT, filter the signal to reduce noise, and set the common-mode ratio voltage of the body. This
circuit was fabricated into a PCB for testing. As large-scale production (a run of 1000 or more), the cost
of the analog front end hardware would only cost approximately $100 for a full turn-key system. This
means that this analog front end is an extremely cost-effective method for ECG signal acquisition.

FIGURE 11 – INPUT BUFFERS AND WILSON CENTRAL TERMINAL

FIGURE 12 - INPUT BUFFER FREQUENCY RESPONSE

17
 FIGURE 13 – OUTPUT STAGE WITH ACTIVE FILTERING

FIGURE 14 - SIMULATED OUTPUT FROM 1MV SINEWAVE @100HZ (500MV/DIV, 50MS/DIV)

18
FIGURE 15 - OUTPUT STAGE FREQUENCY RESPONSE

FIGURE 16 – RIGHT-LEG DRIVE CIRCUIT (RLD)

19
 FIGURE 17 - RIGHT LEG DRIVE FREQUENCY RESPONSE

Step 4 – Embedded Software

FIGURE 18 - EMBEDDED SOFTWARE DIAGRAM

For the embedded board, we will simply be using an ARM Cortex M4 evaluation board from Texas
Instruments that has a sufficient number of analog input and processing power to acquire all 12 analog
signals. The TM4C123G from Texas Instruments provides this functionality as well as a dedicated UART
connection that is used to interface with out SparkFun Bluetooth module. Using these two boards
together provides sufficient data acquisition, processing speed, and wireless output to achieve our
desired results. Using the Bluetooth module, we use Bluetooth Serial (SPP), essentially a wireless serial
connection, to an Android Phone for the data acquisition. Figure 18 shows a basic block diagram of the
embedded software architecture. To comply with AHA recommendations, this system will need to
transfer samples at 500Hz per lead.25 This means a sampling rate of 6kS/s. Assuming that each sample is
12-bits in size, this means that the overall data rate would have to be 72kbps over Bluetooth to achieve
the correct data stream on the smartphone.

25
ibid

20
Step 5 – Android Development

FIGURE 19 - ANDROID SOFTWARE FLOWCHART

For our application, we will be programming for Android using the Java programming language.
Although we were initially going to use Bluetooth LE to communicate with an Apple iPhone, we quickly
realized that Apple places a number of licensing restrictions for Bluetooth peripherals through their
“Made for iPhone” program. We would have had to apply through the program and spend a large sum
of money to get the appropriate licensing so, instead, we had to make the move to Android
smartphones. Another factor was that Android has a very large market share of the smartphone market,
especially in developing countries, so the switch from iPhone to Android had additional benefits and
reasons along with the Bluetooth restrictions.

In order to meet all of the baseline requirements of a 12-lead ECG, a number of complex features were
developed to meet the project goals. The first major challenge was developing software to give a live-
display (cardiac monitor) of all 12 signals being acquired in real-time. In order to do this, data was
received on the smartphone sequentially. Data that was stored in a 12-element array on the embedded
software was sent to the phone over SPP Bluetooth sequentially as the phone requested samples. In
order to comply with AHA regulations for data acquisition for ECGs, the smartphone requests a set of
samples 500 times per second. When the data stream ends, the phone has received the entirely of the
last sample, and can parse the data to the screen accordingly. Furthermore, the last 10 seconds of data
for all 12 leads are stored in a buffer for the ECG “print out.” In order to activate the printout, the user
simply touches an on-screen button.

The most crucial element for functionality was to display the acquired ECG. Using Figure 9 as a guideline,
the application displays the acquired ECG following the aforementioned convention. Additionally, the
user is able to select which of the 12 leads should be shown across the bottom of the image (the 10-
second-long rhythm strip). The last and most important part of the application’s software was to add
functionality to calculate the RR interval (heart rate), PR interval, QT interval, QTc interval, QRS duration,
and axis. In modern digital 12-lead ECGs, the software combines the full 10-seconds of data for all 12

21
leads into a single “mega-lead” for rhythm analysis.26 This allows the software to determine the average
electrical activity from the entire acquisition. The next step was to determine the heart rate. A peak
detection algorithm was developed which was able to distinguish R waves from any other data or noise
in the signal. The internals were then used to determine heart rate. Using the R waves as reference, the
algorithm then searches in both directions until the previous and next R waves to find any other ways in
that space. Using more peak detection, the remainder of the intervals are acquired and their times
converted to milliseconds. In order to determine axis, the algorithm deviates from the “mega-lead” and
simply compares the magnitude of the peaks for Lead I and Lead aVF. The appropriate data is displayed
in a conventional format on the ECG printout. All calculated values were compared with a conventional
12-lead ECG in order to determine clinical significance. Figure 20 shows an example 12-lead ECG
printout from the Android application. Note that the electrode on V1 was coming off due to a poor
application of adhesive on the off-the-shelf electrode. This does not effect the rest of the output.

FIGURE 20 - SAMPLE 12-LEAD ECG OUTPUT

Step 6 – Functionality Test

In order to verify that the 12-lead ECG functions correctly in a clinical environment, we will have to
create a functionality and testing plan. The group currently holds relationships with a number of
cardiologists in the region who are interested in assisting in the testing process. Furthermore, the group
currently possesses a professional 12-lead ECG to use to test equivalency. Before pursuing FDA approval
for the device, we simply have to prove equivalency with pre-existing technology. Our possession of an
ECG will make the verification significantly easier however the process takes a substantial amount of

26
(Clifford, 2006)

22
time that will extend past the scope of this project. In order to test functionality, we plan to have
professionals use the device to determine that it is providing them with all of the information that they
need in order to do their job.

To verify functionality in lab, we will be able to take the ECG cable that would be attached to one of the
group members, acquire an ECG from one device, plug it in to another, and acquire a second. Our
acquisition should be equal in quality, if not better than the professional device. All calculated values for
intervals should be equivalent as well. Printouts should look nearly identical. In order to finalize
verification after the end of the project period, we will likely move on to in-depth clinical trials in order
to test functionality on individuals presenting rhythms which are not simply healthy normal sinus.

Chapter 5 - Project Deliverables

Project deliverables are as follows:

• A working ECG analog front end board that successfully captures all 12 ECG leads with correct
relative magnitude
• An embedded board with software to convert the analog signals to digital signals and transmit
them wirelessly to an Android smartphone over Bluetooth
• A smartphone application designed to acquire the ECG lead signals over Bluetooth and display
them in real time, as well as generate a “print-out” image.

Future development of this project includes creating a fully custom embedded board that matches the
size constraints of the analog front end board for convenient placement into an enclosure. A prototype
board has already been developed by the team, however there is significant debugging to be done until
this can be considered a fully functional replacement for the solution that has already been
implemented. Furthermore, the group intends to pursue a patent application and, after substantial
development and clinical testing, FDA approval.

Chapter 7 - Conclusion
This project has shown to be an extremely viable proof-of-concept that a 12-lead ECG that is accurate,
portable, and low-cost. Considering that retail costs can be as much as 10 times that of hardware costs,
this would mean that our analog solution would cost $1000 or less. Considering that other products on
the market cost well over $8000 do not meet the same level of functionality or convenience that our
mobile-based product provides, we can confidently say that we have developed a cost-effective and
accessible solution for a portable 12-lead ECG. With Bluetooth battery charging and the ability to use
any Android-based smartphone with the device, this project has the ability to appeal to a great number
of audiences in various markets (particularly those where cost of the device is a significant factor).

Although our requirements for proof-of-concept have been met, there are still a number of objectives to
complete as we move forward outside of the scope of the MQP timeline. The group will continue
developing and polishing the software functionality, as well as further iterating over the two PCB designs
that are being used for the project. Furthermore, the device will be continually tested in order to verify
functionality on a number of individuals to further solidify equivalency for potential FDA approval.
Finally, with help of various resources, the group will be compiling a patent application to be submitted

23
to the US Patenting office in order to protect and all intellectual property that has been tested and
developed during the completion of the MQP.

24
Bibliography
Chen, C.-Y., Chang, C.-L., Chang, C.-W., Lai, S.-C., Chien, T.-F., Huang, H.-Y., et al. (2013, March 4). A Low-
Power Bio-Potential Acquisition System with Flexible PDMS Dry Electrodes for Portable Ubiquitous
Healthcare Applications. Retrieved from Sensors: http://www.mdpi.com/1424-8220/13/3/3077/htm

Clifford, G. D. (2006). Advanced Methods & Tools for ECG Data Analysis. Cambridge, MA: Artech House.

DXE Medical. (2015). Retrieved from Physio-Control LifePak 15 Defibrillator/Monitor:

http://www.dxemed.com/physio-control-lifepak-15-defibrillator-
monitor.html?utm_source=google_shopping&gclid=COW0vrTGsMgCFQgHaQodA3cJsw

Emanuel Goldberger, Cardiologist, Dies at 81. (1993, May 12). The New York Times .

iHealth. (2015). http://www.ihealthlabs.com/glucometer/wireless-smart-gluco-monitoring-system/.

Retrieved from iHealth: http://www.ihealthlabs.com/glucometer/wireless-smart-gluco-monitoring-
system/

Johnson, D. (2011, August 3). Physio-Control to put 12-leads in hands of rural systems . Retrieved from
EMS1: https://www.ems1.com/ems-products/Patient-Monitoring/articles/1091036-Physio-Control-to-
put-12-leads-in-hands-of-rural-systems/

Larson, G. (2014, May 1). 12 lead generated sinus rhythm.

Malmivuo, J., & Plonsey, R. (1995). Bioelectromagnetism. Oxford, United Kingdom: Oxford University
Press.

Nimbleheart. (2015). Products. Retrieved from Nimbleheart.com:

http://www.nimbleheart.com/products.htm

Nimunkar, A. J., David Van Sickle, Van Sickle, D., & Webster, J. G. (2009). Low-cost Medical Devices for
Developing Countries. Retrieved from
http://ewh.slc.engr.wisc.edu//publications/conferences/2009/IEEE_EMBC/Medecal_LowCostDevicesIEE
EMBS2009_v14.pdf

Philips. (2015). View details of Philips PageWriter TC70. Retrieved from Philips.com:
http://www.usa.philips.com/healthcare/product/HC860315/pagewriter-tc70-cardiograph

Physio-Control Inc. (2015). Emergency Medical Care Prehospital Products: Monitor/Defibrillators -

LIFEPAK15. Retrieved from Physio-Control Inc.: http://www.physio-
control.com/ProductDetails.aspx?id=2147484979

Rijnbeek, P. R., Kors, J. A., & Witsenburg, M. (2001). Minimum Bandwidth Requirements for Recording of.
Retrieved from American Heart Association: http://circ.ahajournals.org/content/104/25/3087.full.pdf

Texas Instruments. (2013, November). 1.8Vm microPower, CMOS Operational Amplifiers, Zero-Drift
Series. Retrieved from Texas Instruments: http://www.ti.com/lit/ds/symlink/opa333.pdf

25
University of Rochester Medical Center. (2015). Arrhythmias in Children. Retrieved from University of
Rochester Medical Center:
https://www.urmc.rochester.edu/encyclopedia/content.aspx?ContentTypeID=90&ContentID=P01764

Withings. (2015). Withings Pulse. Retrieved from Withings:

http://www.withings.com/us/en/products/pulse

Withings. (2015). Withings Wireless Blood Pressure Monitor. Retrieved from Withings.com:
http://www.withings.com/us/en/products/blood-pressure-monitor

26