Heart Rhythm Analysis using ECG Recorded with a Novel Sternum

based Patch Technology

A Pilot Study

Dorthe B. Saadi

1,2

, Inge Fauerskov

, Armin Osmanagic

, Hussam M. Sheta

, Helge B. D. Sorensen

Kenneth Egstrup

and Karsten Hoppe

DELTA Dansk Elektronik, Lys & Akustik, Venlighedsvej 4, 2970 Hørsholm, Denmark

Electrical Engineering, Technical University of Denmark, Ørsteds Plads, Bldg. 349, 2800 Kgs. Lyngby, Denmark

Department of Medical Research, OUH Svendborg Hospital, Valdemarsgade 53, 5700 Svendborg, Denmark

Keywords: Long-term ECG Monitoring, Home Cardiac Monitoring, Sternum ECG, Patch ECG Recorder.

Abstract: According to the World Health Organization, cardiovascular diseases are the number one cause of death

globally. Early diagnosis and treatment of many of these patients depend on ambulatory electrocardiography

recordings. Therefore a novel wireless patch technology has been designed for easy, reliable long-term ECG

recordings. The device is designed for high compliance and low patient burden. This novel patch technology

is CE approved for ambulatory ECG recording of two ECG channels on the sternum. This paper describes a

clinical pilot study regarding the usefulness of these ECG signals for heart rhythm analysis. A clinical

technician with experience in ECG interpretation selected 200 noise-free 7 seconds ECG segments from 25

different patients. These 200 ECG segments were evaluated by two medical doctors according to their

usefulness for heart rhythm analysis. The first doctor considered 98.5% of the segments useful for rhythm

analysis, whereas the second doctor considered 99.5% of the segments useful for rhythm analysis. The

conclusion of this pilot study indicates that two channel ECG recorded on the sternum is useful for rhythm

analysis and could be used as input to diagnosis together with other clinical tests and medical history.

1 INTRODUCTION

According to the World Health Organization (2013),

cardiovascular diseases (CVDs) are the number one

cause of death globally. They state that CVDs were

responsible for 30% of all deaths in 2008.

CVDs are not only lethal, but they are also

associated with a high economic burden on the

healthcare system. Furthermore, diseases like

ischemic stroke can have high human costs and

decrease significantly the quality of life. Early

diagnosis and treatment of cardiac related diseases is

therefore crucial. For more than hundred years, the

12-lead electrocardiogram (ECG) has served as the

“gold standard” for diagnosis of different heart

conditions, including arrhythmias (Mittal et al.,

2011). The well-chosen and standardized selection

of electrode positions allows a full investigation of

different projections of the electrical activity of the

heart. This allows a careful investigation of the heart

in different “spatial plans”.

However, for some conditions, it is more

important to obtain long-term information about the

general rhythm of the heart from a rhythm analysis.

In this case, an ambulatory long-term ECG

recording is desired. Some examples of conditions

that are not sufficiently managed by baseline 12-lead

ECG recordings are paroxysmal atrial fibrillation

(AF), non-sustained ventricular tachycardia,

unexplained episodes of syncope, and diagnosis of

other cardiac symptoms not explained by a baseline

12-lead resting ECG (Mittal et al., 2011),

(Zimetbaum and Goldman, 2010). It is, however,

important to notice the different possibilities with a

standard 12-lead ECG and an ambulatory recording.

Some of the main advantages of ambulatory ECG

recordings are long monitoring period, detection of

paroxysmal and asymptomatic arrhythmias, remote

monitoring of the patient and correlation between

specific symptoms and the ECG signals.

It is however still important that the ambulatory

long-term ECG recordings have a sufficient quality

for analysis of specific ECG patterns. Some of the

Saadi D., Fauerskov I., Osmanagic A., M. Sheta H., B. D. Sorensen H., Egstrup K. and Hoppe K..

Heart Rhythm Analysis using ECG Recorded with a Novel Sternum based Patch Technology - A Pilot Study.

DOI: 10.5220/0004640900150021

In Proceedings of the International Congress on Cardiovascular Technologies (CARDIOTECHNIX-2013), pages 15-21

ISBN: 978-989-8565-78-5

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

key features in rhythm analysis include the

depolarization of the atria (the P-wave) and the

depolarization of the ventricles (the QRS complex).

An example of a short unfiltered ECG segment

recorded with the ePatch is provided in Figure 1.

Figure 1: Illustration of a normal sinus rhythm ECG

recorded with the novel ePatch technology. The important

ECG markers are indicated in channel 1 for one heart

cycle. The ECG is raw without any digital filtering.

A number of different long-term ambulatory

monitoring techniques are accepted today. One of

the most commonly applied ambulatory ECG

recorders is the HOLTER monitor.

1.1 Traditional Holter Analysis

The HOLTER monitor typically records 2 – 3 ECG

leads continuously for 24 – 48 hours (Zimetbaum

and Goldman, 2010). During a traditional HOLTER

recording, a medical technician attaches the

electrodes and the HOLTER recorder to the patient.

The electrodes are attached to the recorder through

wires. After the monitoring period, the recorder with

the ECG data is returned to the hospital or healthcare

facility. At the hospital, a specially trained nurse

looks through the recorded data using automatic

software and generates a report for the referring

medical doctor. This report contains a general

description of the rhythm during the recording, any

special findings, and a number of descriptive ECG

“snippets” displaying the different rhythms found

during the recording. The report serves as input to

the diagnosis together with other clinical tests and

the medical history of the patient. If the recording is

of sufficient quality, the following parameters may

be determined based on a traditional HOLTER

recording: Average heart rate and heart rate range,

quantification of atrial and ventricular ectopic beats,

and determination of whether AF is present –

including information about pattern of AF initiation

and termination, shortest and longest duration of AF,

heart rate during AF and AF burden (Mittal et al.,

2011).

However, this monitoring technique possesses a

number of disadvantages including cables that affect

the ability to perform some daily activities during

the recording and the lack of real-time data

transmission and analysis. Furthermore, the

relatively short monitoring duration might not be

sufficient for investigation of infrequent arrhythmias

(Zimetbaum and Goldman, 2010), (Rosenberg et al.,

2013). To account for some of these disadvantages,

a novel wireless ECG patch technology was

designed.

1.2 The ePatch Technology Platform

The ePatch heart monitoring platform is designed

according to a “wear and forget” principle. Thus, the

device is designed to be reliable, safe, comfortable

and easy to use for both the patient and the

healthcare professionals. The ePatch is designed as a

technology platform that can be customised to

account for the needs in a high variety of situations.

Some of the advantages and possibilities with this

novel technology platform are listed below:

 Possibility of multi-sensor design with e.g.

accelerometer recordings for activity estimation.

 Splash proof design: Patients can shower while

wearing the ePatch.

 No cables are needed to connect the electrodes to

the recording device. This highly increases the

patient comfort and decreases the patient burden.

 Possibility of wireless data transmission and/or

local data storage. The platform can be adapted

to any desired communication protocol.

 Home monitoring of cardiac patients that might

reduce hospitalizations.

 Possibility of long-term monitoring due to the

expected higher patient comfort and compliance

with wearing the device.

 Module design allows easy adaptation to

different applications.

 Real-time embedded signal processing, e.g.

automatic detection of cardiac events.

In this pilot study, the focus is to investigate the

application of the ePatch for heart rhythm analysis.

The ePatch version applied in this study is CE

approved for 24 hour ambulatory ECG recordings,

and the ECG signals are stored locally on an internal

memory. An illustration of the applied ePatch is

provided in Figure 2. As observed from Figure 2, the

ePatch is placed at the sternum.

CARDIOTECHNIX2013-InternationalCongressonCardiovascularTechnologies

Figure 2: Illustration of the placement of the ePatch

electrode and sensor on the sternum.

The ePatch system consists of two parts: 1) A

bio-compatible, single-use adhesive electrode with

multiple skin contact points that is attached directly

to the skin surface (this part is termed the ePatch

electrode) and 2) A reusable device that contains a

rechargeable battery, electronic parts, data storage

module, equipment for wireless data transmission,

and a signal processing module (this part is termed

the ePatch sensor). The ePatch sensor is attached

directly on the ePatch electrode. This makes the

system completely free of cables. This is designed so

that patients can perform normal daily activities

during the recording. Furthermore, the ePatch is

easily worn under normal clothing and the cable free

design makes it possible for the patient to easily

change clothes during the recording. This patch

design also facilitates a very small and light weight

construction that minimizes the awareness of the

system while wearing it. The two ECG channels are

measured as bipolar derivations from the multiple

skin contact points, cf. Figure 3.

Figure 3: Illustration of the ePatch electrode and sensor

before assembly. The ePatch version applied in this study

records two ECG channels using bipolar derivations from

the multiple skin contact points in the ePatch electrode.

The placement of the ePatch electrode results in

a shorter distance between the bipolar recording

electrodes. This might influence the quality of the

ECG signal. This reservation toward the quality of

ECG signals recorded using patch technologies with

near-field recording electrodes was also stated by

Mittal et al., (2011). However, Rosenberg et al.,

(2013) compared the ability of a patch type ECG

recorder to recognize episodes of AF with a

traditional 24-hours HOLTER recording. They

found an excellent agreement between the patch

recorder and the HOLTER recorder for both AF

episode detection and AF burden estimation during a

24 hour recording in 74 consecutive patients.

The selected electrode placement furthermore

changes the different projections of the cardiac

vector, and hereby changes the appearance of the

ECG slightly. This might induce issues regarding the

medical professional’s ability to recognize different

heart rhythms and hereby reduce the practicability of

the system. On the other hand, the advantages of this

placement are expected to include the benefits of

reduced artefacts from large muscles and large

movements of electrodes and wires. Furthermore,

several studies have shown promising results with

experimental ECG recorders placed at a midsternal

location. Two research groups, (Janata et al., 2008)

and (Lemmert et al., 2011), conducted studies where

7 seconds noise free HOLTER recordings were

visually compared with 7 seconds noise free ECG

recordings from a prototype device developed by

Phillips Healthcare. Janata et al., (2008) investigated

different placements of the experimental device and

compared the ability to recognize the presence of P

waves, PR time, ventricular morphology (QRS

width ≤ 0.12 seconds or prolonged), and rhythm

diagnosis. They found that for the presence of P

waves, PR time and general rhythm diagnosis, the

device location had a significant influence, and that

a midtsternal location was optimal. They generally

obtain a good agreement between the two devices.

Lemmert et al., (2011) investigated the ability to

visually recognize ventricular ectopic beats (VEBs)

and ventricular fibrillation (VF) with ECGs recorded

with the prototype device and EASI lead. The

authors found a very high accuracy between the two

devices for recognition of VEBs and VEB

configuration counts. The results furthermore

showed a perfect agreement between the two devices

for the recognition of VF. The recognition of pace

spikes was better on the standard device.

Furthermore, studies conducted by Puurtinen et al.,

(2009), indicate that, with respect to P-wave

amplitude, the optimal placement of closely spaced

HeartRhythmAnalysisusingECGRecordedwithaNovelSternumbasedPatchTechnology-APilotStudy

bipolar electrodes is diagonally above the standard

12-lead precordial leads V1 and V2.

A review of the literature thus indicates a strong

potential for the recording of relevant ECG signals

on the sternum. However, the described systems are

not completely comparable to the ePatch system and

it is therefore desirable to investigate the practical

usefulness of two ECG channels recorded with the

novel ePatch technology placed at the sternum. This

investigation is thus the focus of this clinical pilot

study. During the study, the analysis of the recorded

ECG signals is performed in a setting fairly similar

to the setting for the traditional HOLTER or

telemetry recordings.

2 METHODS AND DESIGN

This study includes ECG data segments from 25

different hospitalized patients. The choice of

hospitalized patients ensures a realistic amount of

abnormal beat morphologies and abnormal heart

rhythms. Each of the patients was monitored with an

ePatch for approximately 24 hours. All patients were

simultaneously monitored with the regular telemetry

equipment at the hospital department. The ePatch

ECG signals were recorded using a sampling

frequency of 512 Hz and a resolution of 12 bits. In

compliance with IEC 60601-2-47, the ePatch front

end had an analog bandpass filter between 0.67 and

40 Hz. The study was conducted in accordance with

the principles of Good Clinical Practice (GCP)

(Research Ethics Committee ID: S-20120132). All

patients were informed about the study and signed a

written consent form before their inclusion in the

study. The patients were furthermore questioned

about any discomfort and their general satisfaction

with wearing the system. They were asked regarding

their level of satisfaction on an analog scale from

“very satisfied” to “very dissatisfied”. The study

included 15 males and 10 females. The mean Body

Mass Index (BMI) was 27.5 with a standard

deviation of 6.3. The mean age was 71.7 years with

a standard deviation of 13.0 years.

As mentioned, the purpose of the study was to

investigate whether two channel ECG signals

recorded with the ePatch placed at the sternum is

useful for heart rhythm analysis. In a realistic

setting, a medical technician with speciality in

HOLTER or telemetry analysis extracts relevant

ECG segments that are provided to the referring

medical doctor. This step was also performed in this

study. An experienced nurse was asked to extract 7

seconds ECG segments where the interpretation of

the ECG signal was not hindered by noise, in other

words, the data should be of sufficient signal quality.

The definition of sufficient signal quality was thus

based on a subjective judgement by an experienced

ECG analyzer. This is somehow similar to the

extraction of ECG snippets during a traditional

HOLTER analysis. The ECG segments were

provided to both the nurse and the medical doctors

without any form of digital filtering, that is, the

analysis was based on raw ECG signals. A total of 8

segments were extracted from each patient

according to the scheme illustrated in Figure 4.

Figure 4: Illustration of the data extraction and selection

process. The red marks on the top panel indicate the three

one hour segments that were extracted for the study for a

recording of exactly 24 hours. The bottom panel illustrates

the process of selection of 7 seconds segments from the

extracted data. Green segments are selected for the study,

whereas red segments are excluded.

For each patient, three hours of data was

considered. The three hours were extracted as 1 hour

in the beginning, 1 hour in the middle, and 1 hour at

the end of the recording. The first and last 30

minutes were, however, not considered to ensure

that artefacts from mounting and removal of

electrodes did not affect the extracted data. The three

hours of extracted data for a recording of exactly 24

hours is illustrated by red colour marks in Figure 4.

This ensures that a patient is only excluded from the

analysis if the general signal quality is insufficient

throughout the recording.

The 7 seconds ECG segments are extracted from

these three hours of data according to the following

scheme: 1) If the current 7 seconds data segment is

considered noise free, it is selected for the study, and

a new 7 seconds segment is investigated 5 minutes

later. 2) If the current 7 seconds data segment is not

considered noise free, it is excluded from the study,

and a new 7 seconds segment is investigated 1

minute later. 3) If it is not possible to extract 8

segments of sufficient signal quality within these

three hours of data, the patient is excluded from the

study.

The data was selected with a custom designed

Graphical User Interface (GUI) using MATLAB

R2012B. The GUI provided an illustration of 7

seconds of two channel ECG data, and the nurse was

asked to use two check boxes to choose between

CARDIOTECHNIX2013-InternationalCongressonCardiovascularTechnologies

“noise free segment” and “noise disturbed segment”.

The study included a total of 200 two channel ECG

segments.

After selection of the 200 ECG segments, two

medical doctors with experience in ECG

interpretation performed an independent individual

evaluation of each of the ECG segments according

to the usefulness for heart rhythm analysis.

The medical evaluation of each segment was

conducted using another GUI designed in MATLAB

R2012B. This GUI is illustrated in Figure 5. The two

channel ECG signal is visualized and the medical

doctor was asked to choose between two check

boxes, indicating the usefulness of the ECG segment

for rhythm analysis.

3 RESULTS

Each ECG segment was evaluated according to the

usefulness for heart rhythm analysis. The score

“good” indicates that the ECG segment was found

useful for heart rhythm analysis, whereas the score

“bad” indicates that the ECG segment was not

considered useful for rhythm analysis. The

evaluation from both medical doctors is illustrated in

Figure 6.

Table 1: Results from the evaluation for ECG segment

relevance for rhythm analysis from the two medical

doctors.

Segments marked as “good” Number Percentage

Medical doctor 1 197 98.5%

Medical doctor 2 199 99.5%

Both medical doctors 196 98%

At least one medical doctor 200 100%

As observed from Figure 6, the doctors did not

agree on the segments that were not useful for

rhythm analysis. This is also illustrated in Table 1

that contains the percentage of “good” ECG

segments for each doctor, the percentage of ECG

segments considered as “good” by both doctors and

the percentage of segments considered as “good” by

at least one of the doctors.

Of the 25 patients, 22 indicated that they were

very satisfied with wearing the device, 1 indicated to

be satisfied, and 2 did not answer the question.

Furthermore, several patients mentioned that they

did not even notice that they were wearing it.

Figure 5: Illustration of the designed GUI used for the medical doctor evaluation of each 7 seconds ECG segment. The two

channel ECG segment was visualized on a computer screen, and the medical doctor was asked to check one of the check

boxes dependent on his evaluation of the relevance of the current ECG segment for rhythm analysis. Note, that this segment

illustrates a case of AF.

HeartRhythmAnalysisusingECGRecordedwithaNovelSternumbasedPatchTechnology-APilotStudy

Figure 6: Illustration of the evaluation from both medical

doctors for each of the 200 ECG segments. The score “1”

is “good” and indicates that the ECG segment was

considered useful, whereas the score “0” indicates that the

ECG segment was not considered useful for rhythm

analysis.

4 DISCUSSION

Both medical doctors indicated that more than 98%

of the selected ECG segments were diagnostically

meaningful to them, and that the ECG could help

toward a rhythm analysis and diagnosis of the

patient. It should, of course, be stated that the

diagnosis of the patient would contain results from

other relevant clinical tests, medical history, review

of the entire long-term ECG recording, and general

comments from the nurse preparing the ECG report

for the referring medical doctor. The diagnosis is not

imagined to be based solely on the 7 seconds ECG

segments investigated in this study. However, the

results from this pilot study are very promising and

indicate the potential for this novel device for

ambulatory cardiac monitoring.

The fact that the “bad” segments were not the

same for both medical doctors, could indicate a

certain degree of inter reader variability. The

number of doctors could be increased in a future

study to investigate the true inter reader variability.

For the purpose of this pilot study, it is, however,

considered sufficient with the evaluation by two

medical doctors. It should also be stated that even

using the “worst case” of judging all segments

evaluated as “bad” by at least one of the doctors as a

“bad” segment, still results in 98% of the segments

being useful. Furthermore, a traditional HOLTER

recording might also contain segments of data that is

not useful for rhythm analysis. In a real life

situation, cases of doubt about a diagnosis are solved

by discussion and consensus with other doctors. This

is also expected to be the case when ePatch ECG

signals are applied for rhythm analysis.

Another interesting finding is the generally high

patient satisfaction with wearing the system. This is

one of the expected advantages of this novel

technology. The high patient comfort is expected to

allow very long-term monitoring in the future. This

could increase the likelihood of detecting

paroxysmal and infrequent arrhythmias. The higher

patient comfort is also expected to increase the

compliance with wearing the system, and a high

patient compliance is necessary for reliable

monitoring results (Ackermans, Solosko, Spencer,

Gehman, Nammi, Engel and Russell, 2012).

The focus of this pilot study was to obtain

preliminary knowledge about the overall

applicability of ECG signals recorded with the

ePatch on the sternum. Future studies might include

more direct comparisons between ECG signals

recorded synchronously with the ePatch and

traditional HOLTER recordings. Future studies

might also investigate the ability to correctly detect

specific ECG features, e.g. the presence of the P-

wave.

The future possibilities of this type of long-term

ambulatory ECG recorders seem to be very high in

areas like home monitoring, screenings, and follow-

up consultations. However, the knowledge about the

practical application of these new technologies is

still relatively limited due to the lack of large-scale

applications of the technology in everyday clinical

situations. This study contributes to the currently

limited amount of knowledge about the usability of

these patch type ECG recorders. Further

investigations could be conducted to investigate the

usefulness for specific cardiac conditions on a larger

database. Furthermore, in this study, only noise free

ECG segments were presented to the medical

doctors. This mimics the everyday selection of

representative ECG “snippets” for the referring

medical doctor and serves the purpose of this study.

However, large-scale studies should be conducted to

investigate the general level of artefacts and signal

quality with this new patch technology.

5 CONCLUSIONS

This clinical pilot study indicates the medical

usefulness of two channel ECG signals recorded at

the sternum using the novel ePatch technology for

heart rhythm analysis. Furthermore, the 25 included

patients provided positive declarations regarding

CARDIOTECHNIX2013-InternationalCongressonCardiovascularTechnologies

their experience with the device. Further studies

should be conducted to establish possible new

application areas for this new technology and to

determine the general quality of the signal and the

vulnerability to different types of artefacts.

ACKNOWLEDGEMENTS

The clinical study was supported by funding from

the Danish Business Innovation Fund. The authors

wish to thank the clinical staff at the Department of

Medical Research, OUH Svendborg hospital, for

conducting the clinical recordings.

REFERENCES

Ackermans, P. A. J., Solosko, T. A., Spencer, E. C.,

Gehman, S. E., Nammi, K., Engel, J. & Russell, J. K.

(2012). A user-friendly integrated monitor-adhesive

patch for long-term ambulatory electrocardiogram

monitoring. Journal of Electrocardiology, 45(2), 148-

153. doi:10.1016/j.jelectrocard.2011.10.007.

IEC 60601-2-47:2012 Medical electrical equipment – Part

2-47: Particular requirements for the basic safety and

essential performance of ambulatory

electrocardiographic systems.

Janata, A., Lemmert, M. E., Russell, J. K., Gehman, S.,

Fleischhackl, R., Robak, O., Pernicka, E., Sterz, F. &

Gorgels, A. P. M. (2008). Quality of ECG Monitoring

with a Miniature ECG Recorder. Pacing and Clinical

Electrophysiology, 31(6), 676-684. doi:

10.1111/j.1540-8159.2008.01070.x.

Lemmert, M. E., Janata, A., Erkens, P., Russell, J. K.,

Gehman, S., Nammi, K., Crijns, H. J. G. M., Sterz, F.

& Gorgels, A. P. M. (2011). Detection of ventricular

ectopy by a novel miniature electrocardiogram

recorder. Journal of Electrocardiology, 44(2), 222-

228. doi: 10.1016/j.jelectrocard.2010.10.028

Mittal, S., Movsowitz, C. & Steinberg, J. S. (2011).

Ambulatory External Electrocardiographic Monitoring

Focus on Atrial Fibrillation. Journal of American

College of Cardiolog, 58(17), 1741-1749. doi:

10.1016/j.jacc.2011.07.026.

Puurtinen, M., Viik, J. & Hyttinen, J. (2009). Best

Electrode Locations for a Small Bipolar ECG Device:

Signal Strength Analysis on Clinical Data. Annals of

Biomedical Engineering, 37(2), 331-336. doi:

10.1007/s10439-008-9604-y.

Rosenberg, M. A., Samuel, M., Thosani, A. & Zimetbaum,

P. J. (2013). Use of a Noninvasive Continuous

Monitoring Device in the Management of Atrial

Fibrillation: A Pilot Study. Pacing and Clinical

Electrophysiology, 36(3), 328-333. doi:

10.1111/pace.12053.

World Health Organization (March 2013). Cardiovascular

diseases (CVDs). Retrieved June 6, 2013, from

http://www.who.int/mediacentre/factsheets/fs317/en/

Zimetbaum, P. & Goldman, A. (2010). Ambulatory

Arrhythmia Monitoring: Choosing the Right Device.

Circulation, 122(16), 1629-1636. doi:

10.1161/CIRCULATIONAHA.109.925610.

HeartRhythmAnalysisusingECGRecordedwithaNovelSternumbasedPatchTechnology-APilotStudy