Prediction of Local Abnormal Ventricular Myocardial Electrical

Activation on Surface ECG in Patients with Structural Heart Disease

Zafar M. Yuldashev

, Anatoli P. Nemirko

, Evgeny N. Mikhaylov

, Dmitry S. Lebedev

Aleksei A. Anisimov

, Alena I. Skorobogatova

and Darina S. Ripka

Saint Petersburg Electrotechnical University, Professor Popov Street 5, Saint-Petersburg, Russia

Almazov National Medical Research Centre, Akkuratov Street 2, Saint-Petersburg, Russia

Keywords: Endo- and Epicardial Electrograms, Surface ECG, Synchronous Registration, Signal Processing and Analysis,

Ventricular Late Potentials, Correlation, Life-threatening Heart Disorders.

Abstract: The problems of processing and analysis synchronous records endo- and epicardial electrograms and surface

ECG signals, detection of ventricular late potentials of patients with ventricular tachyarrhythmia and coronary

heart disease by using spatial and temporal signal accumulation algorithms, correlation of temporal and

spectral characteristics of late potentials with the dynamics and localization of dangerous heart disorders are

considered.

1 INTRODUCTION

The surface ECG signal is the result of the spatio-

temporal summation of the electric potential, which

is formed as a result of the excitation of myocardial

fibres by the action potential when it spreads along

the conduction pathways of the heart and the

contractile myocardium. Normally, the action

potential is first generated by the sinoatrial node and

sets the heart rate. A cardiac electrogram is an

electrical signal recorded by a pair of electrodes of a

special catheter when the action potential passes by

this pair of electrodes. The temporal characteristics

and form of intracardiac electrograms on paired

catheter electrodes represent the nature of the

propagation of the action potential along the

myocardium. There is a relation between the

characteristics of intracardiac electrograms and the

waves and segments of the surface ECG signal. Endo-

and epicardial electrogram registration is of

paramount importance in diagnosis and treatment of

heart rhythm disorders, since they allow establishing

https://orcid.org/0000-0003-1075-3420

https://orcid.org/0000-0001-6459-626X

https://orcid.org/0000-0002-6553-9141

https://orcid.org/0000-0002-2334-1663

https://orcid.org/0000-0003-1363-1971

https://orcid.org/0000-0002-5490-6217

https://orcid.org/0000-0002-0244-4487

the localization and mechanism of tachyarrhythmias

and conduction abnormalities. Ventricular late

potentials and fragmented QRS complexes have

significant diagnostic value. A large number of

publications that have become classics of clinical

cardiology (Breithardt, Borggrefe, Martinez-Rubio,

et al, 1988; Simson, Euter, Michelson et al, 1981;

Simson, 1983; Breithardt, Cain, El-Sherif et al., 1991)

show that these potentials are reliable predictors of a

number of life-threatening heart disorders and sudden

death. However, the registration and analysis of endo-

and epicardial electrograms is carried out only for

certain indications, in contrast to the registration and

analysis of the surface ECG signal, the identification

and evaluation of the characteristics of ventricular

late potentials and fragmented potentials is difficult

due to their short duration (less than 180 ms), low

amplitude (less than 40 μV), significant frequency

variability (up to 700 Hz), amplitude and duration.

Since in the majority cases ventricular fragmented

and late potentials are not detected on surface ECG

using simle analysis, special methods and algorithms

Yuldashev, Z., Nemirko, A., Mikhaylov, E., Lebedev, D., Anisimov, A., Skorobogatova, A. and Ripka, D.

Prediction of Local Abnormal Ventricular Myocardial Electrical Activation on Surface ECG in Patients with Structural Heart Disease.

DOI: 10.5220/0009374103950401

In Proceedings of the 13th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2020) - Volume 1: BIODEVICES, pages 395-401

ISBN: 978-989-758-398-8; ISSN: 2184-4305

395

for processing and analyzing multi-channel

recordings of a surface ECG signal are required.

In patients with structural heart diseases

inhomogeneous and delayed electrical activation is

associated with re-entrant and triggered life-

threatening ventricular tachyarrhythmias, as it had

been shown in many scientific researches (Kreiner,

Gottlieb, Furukawa et al., 1992; Wong and Windle,

1994; Teptin, Latfullin, Kоnturov, Mamedova, 2004;

Ohisa, Ohira, Mizonobe et al., 2002). Usually, local

abnormal electrical activation of ventricular

myocardium can be recorded invasively only, when a

mapping catheter is placed at endocardial or

epicardial surface in close proximity to the diseased

area. In clinical practice, invasive mapping of

ventricular tachycardia substrate is aimed at detection

of areas with low amplitude and abnormal local

activity. Although the presence and location of

abnormal activity can be predicted according to

patient clinical characteristics (for instance, in

patients with known localization of post-myocardial

infarction scar), there is a need in prediction models

of the presence and extent of abnormal activity areas

in patients with other cardiac diseases. Pre-procedure

knowledge of localization of the target area for

mapping and further catheter ablation (in order to

terminate and render VTs non-inducibleis of

paramount importance, since it helps to plan the

required access and needed extent of tissue ablation.

The purpose of this study is to develop a

processing algorithm for detecting ventricular late

potentials and fragmented signals from synchronous

recordings of surface ECG and invasively registered

signals, and to improve the accuracy of estimation the

presence of local abnormal electrograms and their

spectral characteristics using the analysis of surface

ECG.

To achieve this goal, we have solved the

following research tasks:

1. Formation of a database of synchronous recordings

of endo- and epicardial electrograms and 12-channel

surface ECG signals reflecting the presence and

absence ventricular late potentials and fragmented

electrograms for various heart rhythm disorders.

2. Detection, analysis and classification of ventricular

late potentials and fragmented QRS complexes using

algorithms for synchronous accumulation and spatial

averaging over surface ECG signals, comparison of

the accuracy of detection of ventricular late and

fragmented potentials and assessment of their

characteristics taking into account synchronous

recordings of intracardiac electrograms.

3. Formation of a complex of indicators of surface

ECG signals correlating with intracardiac ventricular

fragmented and late potentials.

4. Development of an algorithm for identifying late

and fragmented ventricular potentials and evaluating

their temporal, spectral and dynamic characteristics.

5. Formation of a complex of indicators of

fragmented QRS complexes and ventricular late

potentials, reflecting dangerous heart rhythm

disorders.

2 METHODS

2.1

Patient Population

Patients with known structural heart disease and

documented ventricular tachycardia (VT) were

referred for electrophysiological study and catheter

ablation of VT substrate. Inclusion criteria were the

following: the presence of structural myocardial

disease diagnosed using transthoracic

echocardiography, magnetic resonance tomography,

and/or endomyocardial biopsy; VT detected on

surface ECG or by interrogation of an existent cardiac

implantable electronic device (mainly, implantable

cardioverter-defibrillator); signed informed consent

to undergo an invasive electrophysiological study.

Exclusion criteria were the following: the presence of

a reversible VT cause, acute systemic inflammatory

disease, intracardiac thrombosis, the need for

coronary revascularization according to the clinical

and angiographic evaluation.

2.2

Electrophysiological Study

The electrophysiological procedure was performed in

an electrophysiological laboratory; patients were

evaluated in a fasting state under general anesthesia

with propofol, fentanyl and arduan. A femoral access

was performed via the common femoral vein (a

transseptal 8F Multipurpose sheath (Cordis, Johnson

and Johnson, USA) and a 6F vascular sheath (Avanti,

Cordis, Johnson and Johnson, USA) were

introduced), and via the common femoral artery (an

8F vascular sheath was used).

A combined endocardial left ventricular access

was performed retrogradely via the arterial sheath and

using the transseptal access. Puncture of the

interatrial septum was performed using the

Brockenbrough BRK-1 needle (Abbott, USA) with a

small amount of contrast media used to confirm

appropriate access to the left atrium (Optiray 300,

Mallickrodt, Germany).

After successful transseptal puncture the

transseptal sheath was advanced into the left

NDNSNT 2020 - Special Session on Non-invasive Diagnosis and Neuro-stimulation in Neurorehabilitation Tasks

396

ventricle. Following left-sided access intravenous

heparin was administered (100 IU* kg-1) to prevent

thrombosis. A 6F quadripolar diagnostic catheter

(Webster, Johnson and Johnson, USA) was placed

into the right ventricle apex for stimulation and/or

time annotation of the bipolar intracardiac signals.

Electrogram mapping was performed using both a

duodecapolar Pentaray catheter (Biosense Webster,

USA) and a 3.5-mm tip quadripolar irrigated mapping

and ablation catheter NaviStar ThermoCool

(Biosense Webster, USA). Electrophysiological

mapping was performed under the non-fluoroscopic

three-dimensional mapping system CARTO 3

(Biosense Webster, USA).

Three-dimensional shells were created using the

“FAM (Fast Anatomical Model)” module and

automatic point acquirement using the “Confidense”

module with a maximum 2 mm distance between

points. Electrophysiological signals from mapping

catheters were recorded and stored simultaneously

with surface ECG signals on the CardioLab (GE,

USA) system.

When epicardial mapping was indicated and

planned, fluoroscopically-guided subxyphoid

puncture was performed first; the technique was

previously described in details (Simonova, Lebedev,

Mikhaylov, 2017; Simonova, Mikhaylov, Tatarskiy

et al., 2019). The non-steerable 8F multipurpose

sheath was inserted into pericardial space for

introducing and manipulating a mapping catheter.

Endo- and epicardial mapping was performed

during sinus rhythm or during right ventricle

stimulation at a rate 600 ms per min. Normal bipolar

signals were characterized by two high-frequency

deflections (a positive and a negative consecutive

deflections). Abnormal signals were characterized by

splitting, notching, slurring, fragmentation, doubling

(the presence of an isoline between two components),

and by the late activity (signals widely separated from

the main signals and located after the end of QRS on

the surface ECG).

2.3

Electrophysiological Signals’

Extraction

Tracings with normal and abnormal bipolar signals

were manually annotated on the electrophysiological

system, and tracings containing the ECG and

intracardiac/epicardial bipolar signals were extracted

from the electrophysiological system in .txt format.

The extraction was performed using an integrated

module which allowed marking the cut-off timings on

the whole registration. The raw signal was recorded

at 1000 Hz sampling rate.

3 PROBLEMS SOLVING

3.1

Forming of the Synchronous

Records Base of Endocardial

Electrograms and Surface ECG

Signal

After selecting the most informative segments of the

signals and storing them, the logic of constructing a

database of synchronous records was formulated: a

hierarchical structure of the presentation was chosen.

Its use makes it easy to find records of interest by type

of heart rhythm disturbance and to reveal the

dynamics of the parameters before and after invasive

treatment (radiofrequency catheter ablation of a

tachycardia critical isthmus and/or areas with local

abnormal electrical potentials).

The first level of division is the type of observed

ventricular disturbance: fibrillation, flutter,

tachycardia, bradycardia, extrasystole and late

potentials (Yuldashev, Nemirko, Manilo et al., 2019).

After dividing the recordings by type of

disturbance, the signals were divided into 3 additional

levels: recordings before RF ablation, during and

after the ablation. Using this division allows to track

the change in the characteristics of the electrical heart

activity throughout the operation.

Surface ECG signals were recorded in three or

twelve leads. Signal record database contains 3

channels of surface ECG: I, II and III. For disorders

associated with the ventricles, all 12 ECG channels

were used. Intracardiac activity is represented by the

following leads:

1) dABL – distal lead of the ablation catheter

electrode;

2) ABL – proximal lead of the ablation catheter

electrode;

3) CS12, CS34, CS 56, CS78, CS910 – leads of the

catheter electrode located in the coronary sinus;

4) RV – signals from distal bipoles of the diagnostic

catheter placed in the right ventricle apex;

5) Pentaray – bipolar signals recorded from the

duodecapolar steerable mapping catheter roving in

and on the ventricles.

The database of records consists of 296 records

lasting from 10 to 20 minutes, including records of

the norm and atrial pathology – 128 records, various

ventricular pathologies – 168 records.

Prediction of Local Abnormal Ventricular Myocardial Electrical Activation on Surface ECG in Patients with Structural Heart Disease

397

3.2

FQRS Complex and Ventricular

Late Potentials Detection

Algorithms

The following algorithm was proposed to detect

fragmented QRS (FQRS) complexes (figure 1). At

the first stage, the surface ECG signal was subjected

to low-pass filtering with a cutoff frequency of the

filter fCH = 100 Hz to reduce the influence of high-

frequency noise during the extraction of QRS

complexes.

Figure 1: Fragmented QRS detection algorithm.

In the reason of the fragmented QRS complex is

characterized by the presence of notches, patterns in

the areas of Q, R and S waves (figure 2), to detect

them, the initial ECG signal in the segment of the

selected QRS complex undergoes differentiation.

Next, the irregularity of the differentiated signal is

established and its analysis. A smooth change in the

differentiated signal from a negative value to a

positive value means the absence of fragmented QRS

complexes. Its stepwise change reflects the presence

of fragmented complexes.

Methods of temporal or spatial summation are

used to detect ventricular late potentials. It should be

noted that the ventricular late potentials appear on the

ST segment and are characterized by a very low

amplitude (tens of μV), high frequency (up to

hundreds of Hz), short duration (up to 150 ms) and

significant variability of characteristics from one

cardiocycle to another. This fact makes it difficult to

detect and analyze ventricular late potentials (VLP).

Figure 2: Types of fragmented QRS complex.

Methods of temporal or spatial summation are

used to detect ventricular late potentials (figure 3). It

should be noted that the ventricular late potentials

appear on the ST segment and are characterized by a

very low amplitude (tens of μV), high frequency (up

to hundreds of Hz), short duration (up to 150 ms) and

significant variability of characteristics from one

Figure 3: Temporal summation VLP detection method.

NDNSNT 2020 - Special Session on Non-invasive Diagnosis and Neuro-stimulation in Neurorehabilitation Tasks

398

cardiocycle to another. This fact makes it difficult to

detect and analyze ventricular late potentials (VLP).

When using temporary summation (figure 3), R

prongs are first allocated on the ECG signal, relative

to which a window that coincides with the ST

segment is formed. Within this window, for all N

(usually up to 150) cardiocycles, synchronous

accumulation (summation) of discrete ECG signal

counts is performed.

For one cardiocycle, the number of samples can

reach m = 200. With synchronous accumulation, the

amplitude of the late ventricular potential increases N

times and reaches a level of tens of mV. It is filtered

by a band-pass filter in the range from 100 to 250 Hz

in order to eliminate the low-frequency components

of the ST segment. The rest of the signal represents

the ventricular late potential. The considered method

of synchronous signal accumulation for detecting the

ventricular late potential has advantages and

disadvantages. The advantage is the simplicity and

using only one channel of the surface ECG signal.

Disadvantages exceed advantages. The presence of

noise fluctuations in the ECG signal leads to the

detection of R prongs and the formation of a window

within which synchronous accumulation is

performed, with an error of up to 2-3 reports of the

sampling signal at a sampling frequency of 1,0 kHz,

which in turn smears the ventricular late potential and

distorts the high-frequency components. Another

disadvantage is the inability to assess the dynamics of

the characteristics of the VLP and their duration due

to the long stage of accumulation (up to 150

cardiocycles).

The spatial accumulation method (figure 4) is

devoid of these disadvantages. However, it requires

performing synchronous recordings of surface ECG

signals over 12 channels, detecting R prongs, forming

the window for detecting the samples of ECG signals

on ST segment, summing up identical discrete

samples across all channels, band-pass filtering of the

resulting signal in the range from 100 to 250 Hz.

The resulting signal represents the ventricular late

potential obtained from the spatial summation of

multichannel ECG signals. The advantage of the

considered method is that it allows reflecting the

dynamics of the characteristics of the VLP signal, and

is less sensitive to fluctuation noise, because the

formation of the window and synchronization of the

summation of the samples is carried out

simultaneously on all channels. However, this

method has a drawback. It does not allow to

significantly increase the amplitude of the VLP

signal, the gain does not exceed 12.

Figure 4: Spatial accumulation VLP detection method.

3.3 Fragmented QRS and VLP Research

Studies of ventricular late potentials were performed

on 16 patients using synchronous recording of

intracardiac ventricular electrograms and 12 leads of

surface ECG (Yuldashev, Anisimov, Nemirko et al.,

2019). The ventricular late potentials were detected

on intracardiac electrograms in all cases, while

surface ECGs did not visually reveal these

abnormalities due to very low levels of late potentials

amplitude, short duration, and high frequency of

electrical vibrations (figures 5 -7).

The ventricular late potentials detected by

intracardiac electrograms almost in most cases

correlate well with various ventricular myocardial

disorders. As the results of studies [1, 4, 6, 7, 8] show,

the accuracy and sensitivity of the diagnosis of both

Prediction of Local Abnormal Ventricular Myocardial Electrical Activation on Surface ECG in Patients with Structural Heart Disease

399

Figure 5: Example of VLP detection.

Figure 6: Example of VLP detection.

Figure 7: Example of VLP detection.

dangerous cardiac arrhythmias due to cardiac

conduction disorders and myocardial cell

morphology using ventricular late potentials is at least

92%.

The results of these studies confirm the need for

further improvement of methods and technologies for

identifying fragmented and late potentials using

surface ECGs and verifying their results using

intracardiac electrograms.

3.4 Clinical and Development

Perspective

To the best of our knowledge, the signal database

created within the scope of this work is one of the first

of its kind and will be used in future research.

Processing algorithms for automatic local

electrical abnormal potentials detection are under

development in this project, and, once developed, will

be useful with potential future implementation into

invasive electrophysiological diagnostic systems.

The prediction algorithms based on surface ECG

analysis could be useful in estimation of the presence

and localization of local abnormal electrical activity

and will be of clinical importance, since might be

implemented into pre-procedure planning of the

access and extent of catheter ablation.

4 CONCLUSIONS

The results of the studies confirm the conclusions

about the need and feasibility of using the ventricular

late potentials and fragmented potentials for the

diagnosis of dangerous heart disorders. Of course, in

a clinical setting, the results of recording intracardiac

electrograms can be used to diagnose that disorders.

However, often there is a need for the diagnosis of

cardiac abnormalities outside the clinic, in particular

at home using a wide range of electrocardiographs.

To diagnose heart disorders that pose a threat to the

patient’s life, at home it is necessary to use tools and

software that will record surface ECG signals and

identify fragmented and late heart potentials hidden

in surface ECG signals. Given that the accuracy of the

diagnosis of such heart disorders using methods of

pre-processing and synchronous signal accumulation

is quite high, such devices will significantly improve

the quality of medical care for cardiac patients.

ACKNOWLEDGEMENTS

This research was supported by Russian Foundation

for Basic Research (RFBR), research projects 18-29-

02036, 19-29-01009.

REFERENCES

Breithardt, G., Borggrefe, M., Martinez-Rubio, A. et al,

1988. Prognostic significance of ventricular late

NDNSNT 2020 - Special Session on Non-invasive Diagnosis and Neuro-stimulation in Neurorehabilitation Tasks

400

potentials in the postmyocardial infarction period. In

Herz. Bd 13, №3, pp.180-187.

Simson, М. В., Euter, D. E., Michelson E. L., et al, 1981.

Detection of delayed ventricular activation on the body

surface in dogs. In Amer. J. Physiol. Vol. 241, pp.

H363-H369.

Simson, M. B., 1983. Clinical application of signal

averaging. In Cardiol. Clin. Vol. 1, pp. 109-190.

Breithardt, G., Cain, M. E., El-Sherif, N. et al., 1991.

Standards for analysis of ventricular late potentials

using high resolution or signal-averaged

electrocardiography. A statement by a Task Force

Committee between the European Society of

Cardiology, the American Heart Association and the

American College of Cardiology. In Eur. Heart.

Journal. Vol. 12(4), pp. 473-480.

Kreiner, G., Gottlieb, C. D., Furukawa, S. et al., 1992. Late

potentials in an ovine model of acute transmural

myocardial infarction. In J. Appl. Physiol. Vol. 73(3),

pp. 841-846.

Wong, C. B., Windle, J. R., 1994. Clinical applications of

signal averaged electrocardiography in patients after

myo cardial infarction. In Nebraska Medical Journal.

Vol. 79(2), pp. 28-31.

Teptin, G. M., Latfullin, I. A., Kоnturov, S. W., Mamedova,

L. E., 2004. Analysis of low amplitude cardiac signals

and its interpretation. In Environmental Radioecology

and applied ecology. Vol. 10, №1, pp. 3-7.

Ohisa, N., Ohira, M., Mizonobe, K. et al., 2002.

Comparative study of T-wave alternans, QT c

dispersion and late potential for predicting ventricular

tachycardia in patients with ischemic heart disease. In

Rinsho Byori. Vol. 50(2), pp.191-195.

Simonova, K. A, Lebedev, D. S, Mikhaylov, E. N., 2017.

Epicardial mapping and ablation in management of

ventricular tachycardia. In Complex Issues of

Cardiovascular Diseases. Vol 6(4), pp. 138-145.

Simonova, K. A, Mikhaylov, E. N, Tatarskiy, R. B et al.,

2019. Epicardial arrhythmogenic substrate in patients

with postinfarction ventricular tachycardia: a pilot

study. In Journal of Arrhythmology. Vol 1(95), pp. 38-

46.

Yuldashev, Z., Nemirko, A., Manilo, L. et al., 2019.

Processing of Synchronous Recordings of Surface ECG

and Intracardiac Potentials for Diagnostics of

Dangerous Heart Rate Disturbances. In Proceedings

Ural Symposium on Biomedical Engineering,

Radioelectronics and Information Technology

(USBEREIT). PP. 102-105.

Yuldashev, Z., Anisimov, A., Nemirko, A. et al., 2019.

Heart rate disorders identification using superficial

ECG and invasive electrogram. In AIP Conference

Proceedings.

Prediction of Local Abnormal Ventricular Myocardial Electrical Activation on Surface ECG in Patients with Structural Heart Disease

401