Preface

**Section 3 Advances in Endocardial Signal Processing 135**

**Expression of Patient Outcomes 169**

Becerra and Javier Saiz

Avila

**VI** Contents

Chapter 8 **Complexity of Atrial Fibrillation Electrograms Through**

**Nonlinear Signal Analysis: In Silico Approach 137**

Chapter 9 **Loss of Complexity of the Cardiac Bioelectrical Signal as an**

Catalina Tobón, Andrés Orozco‐Duque, Juan P. Ugarte, Miguel

Pedro Eduardo Alvarado Rubio, Ricardo Mansilla Corona, Lizette Segura Vimbela, Alejandro González Mora, Roberto Brugada Molina, Cesar Augusto González López and Laura Yavarik Alvarado

Firstly, I want to thank each of the authors for their contributions to this interesting and unique book project.

It certainly represents a tour de force and expansive review of cardiac electrogram interpre‐ tation intended for the advanced reader but highlighting certain basic concepts as well. These individual sections represent both works that are directly derived from clinical stud‐ ies and relevant anecdotal cases as well as chapters on clinically relevant basic science.

Individual manuscripts have been compiled and edited to constitute the chapters. Each chapter tries to maintain the focus on interpreting the studies and data in a clinically rele‐ vant format that can be translated to patient care. The latter chapters in this body of work focus on the potential of innovative electrograms that have been mathematically derived and correlated to patient outcomes.

The first section examines the surface ECG. This is by no means a basic review of an ap‐ proach to ECG readings but rather assumes that the reader is at an advanced level and starts off by introducing a diagnostic algorithm that Dr. Chacko et al. propose for diagnosing su‐ praventricular tachycardias.

Left axis deviation on the ECG is explored by Dr. Madhur.

I am pleased to also include a chapter that addresses the diagnostic potential of the signalaveraged ECG.

This highly useful surface of electrogram derivation has certainly been neglected by general physicians as well as cardiologists alike and is a useful way of amplifying low amplitude cardiac potentials that can explain the mechanism or assess the potential risk of ventricular arrhythmias. This manuscript by Ioana et al. starts by reinforcing the diagnostic criteria of the signal-averaged ECG and thereafter explores its application.

The contribution by Dr. Xin Gao looks specifically at the fetal ECG and the filtering required in order to discern more clearly fetal heart rate monitoring. This potential may lead to the detection of fetal arrhythmias in utero.

In the second section of this book, manuscripts that interpret electrograms from intracardiac pacemakers and implantable cardioverter-defibrillators (ICDs) are examined in closer detail.

Both Dr. Erik Wissner et al. and Dr. Sadiq Ali et al. look at the interpretation of these device electrograms for the purposes of diagnosis and troubleshooting of ICDs, and Dr. Cismaru Gabriel et al. specifically examine the intracardiac electrograms from ICDs implanted in pa‐ tients with Brugada syndrome. This section is not only relevant for the elimination of inap‐ propriate therapies, which still plague patients with ICDs and various studies quote incidences of up to 16%, but also addresses the concept of a more standardized program‐ ming for both primary and secondary indications.

The concept of mere "shock box" single-zone programming for primary prevention is chal‐ lenged and summarized by examining all recent randomized controlled trials. The strategy proposed is to employ multiple VT detection zones and apply advanced detection algo‐ rithms in all zones given the advancement and high specificity of these algorithms, which include morphological discriminators in most devices.

The section on the Brugada syndrome examines specifically the morphological interpreta‐ tion of the specific intracardiac signals during ventricular arrhythmias in this disease.

This is brought to clinical relevance by proposing to use the stored ICD events and electro‐ grams (EGMs) to plan and guide mapping and catheter ablation both in the case of mono‐ morphic re-entry ventricular tachycardia and in the case of Brugada syndrome, ablation of PVC triggers that initiate VF. Each of these concepts is well illustrated with case studies and actual patient intracardiac electrograms (EGMs).

The final section in this book looks at various basic science concepts in a clinically applied context. Catalina Tobón et al. review complex fractionation from intracardiac EGM sources using nonlinear signal analysis. The concept of targeting complex fractionated electrograms (CFE) during ablation for persistent and long-standing persistent atrial fibrillation (AF) has been controversial. More recently, ablation of cardiac ganglia, specific optical mapping and ablation AF rotors have entered the arena. Various studies have been proposed that examine atrial fibrillation CFE signals in order to see if they exhibit temporospatial stability and if these signals can be used as a surrogate marker to identify rotors.

The final manuscript in this book proposes a controversial concept of using a bioelectrical signal derived from surface electrocardiograms in patients to predict the prognosis in pa‐ tients. It is not a replacement for the 12 lead ECGs but is an interesting additional concept that has been explored in other studies. The data are at best anecdotal and need refinement, but nevertheless warrant closer scrutiny.

In summary, this book as a collation of manuscripts incorporates clinical data and basic sci‐ ence concepts in relation to cardiac signal processing and interpretation.

> **Kevin A. Michael, MBChB, MPhil, MD** Associate Professor, Department of Medicine, Queen's University Kingston, Ontario, Canada

**Concepts from the Surface Electrocardiogram**

incidences of up to 16%, but also addresses the concept of a more standardized program‐

The concept of mere "shock box" single-zone programming for primary prevention is chal‐ lenged and summarized by examining all recent randomized controlled trials. The strategy proposed is to employ multiple VT detection zones and apply advanced detection algo‐ rithms in all zones given the advancement and high specificity of these algorithms, which

The section on the Brugada syndrome examines specifically the morphological interpreta‐ tion of the specific intracardiac signals during ventricular arrhythmias in this disease.

This is brought to clinical relevance by proposing to use the stored ICD events and electro‐ grams (EGMs) to plan and guide mapping and catheter ablation both in the case of mono‐ morphic re-entry ventricular tachycardia and in the case of Brugada syndrome, ablation of PVC triggers that initiate VF. Each of these concepts is well illustrated with case studies and

The final section in this book looks at various basic science concepts in a clinically applied context. Catalina Tobón et al. review complex fractionation from intracardiac EGM sources using nonlinear signal analysis. The concept of targeting complex fractionated electrograms (CFE) during ablation for persistent and long-standing persistent atrial fibrillation (AF) has been controversial. More recently, ablation of cardiac ganglia, specific optical mapping and ablation AF rotors have entered the arena. Various studies have been proposed that examine atrial fibrillation CFE signals in order to see if they exhibit temporospatial stability and if

The final manuscript in this book proposes a controversial concept of using a bioelectrical signal derived from surface electrocardiograms in patients to predict the prognosis in pa‐ tients. It is not a replacement for the 12 lead ECGs but is an interesting additional concept that has been explored in other studies. The data are at best anecdotal and need refinement,

In summary, this book as a collation of manuscripts incorporates clinical data and basic sci‐

**Kevin A. Michael, MBChB, MPhil, MD**

Kingston, Ontario, Canada

Associate Professor, Department of Medicine, Queen's University

ming for both primary and secondary indications.

VIII Preface

include morphological discriminators in most devices.

actual patient intracardiac electrograms (EGMs).

but nevertheless warrant closer scrutiny.

these signals can be used as a surrogate marker to identify rotors.

ence concepts in relation to cardiac signal processing and interpretation.

**Provisional chapter**

## **An Approach to Diagnosing Supraventricular Tachycardias on the 12-Lead ECG Tachycardias on the 12-Lead ECG**

**An Approach to Diagnosing Supraventricular** 

DOI: 10.5772/intechopen.70143

Sanoj Chacko and Adrian Baranchuk Sanoj Chacko and Adrian Baranchuk Additional information is available at the end of the chapter

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/intechopen.70143

#### **Abstract**

Supraventricular tachycardia (SVT) is a general term describing a group of arrhythmias whose mechanism involves the atria and atrioventricular nodal tissue for its initiation and maintenance. SVT is a common entity in clinical practice with a prevalence of 2.25 cases per 1000 in general population. Atrial fibrillation and atrial flutter are the most common presentations of SVTs. Of the remaining subtypes of SVT, atrioventricular nodal re-entrant tachycardia (AVNRT) accounts for 60% of the cases. The atrioventricular reentrant tachycardia (AVRT) and atrial tachycardia (AT) represent approximately 30 and 10% of the cases, respectively. The mechanisms of different forms of SVT have been elucidated and are caused by either re-entrant circuit, increased automaticity or triggered activity. This chapter provides an overview of how to systematically approach a narrow complex tachycardia.

**Keywords:** supraventricular tachycardia, atrioventricular nodal re-entry tachycardia, atrioventricular re-entry tachycardia, atrial tachycardia, atrial fibrillation, atrial flutter

## **1. Introduction**

A systematic approach in the interpretation of ECG is important to arrive at a definitive diagnosis of the subtype of supraventricular tachycardia (SVT). Narrow complex tachycardias (NCT) are broadly divided as regular or irregular tachycardia. Irregular NCT are either atrial fibrillation or atrial flutter with variable ventricular response. For a regular NCT, the next key step would be to identify the presence of a P wave, positive or negative P waves, and its morphology and location of the P wave with respect to the cardiac cycle. Absence of P wave in a regular NCT narrows the differential diagnosis to either typical atrioventricular nodal re-entrant tachycardia (AVNRT) or junctional tachycardia. If the P waves are positive and are located before the QRS complex, it is likely to be sinus tachycardia. On the other

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2017 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

hand, if the P waves are negative, that excludes a sinus origin of the rhythm. A negative P wave in front of the QRS complex implies a low atrial tachycardia (LAT). If the P waves embedded in the QRS complex have a short RP interval, it is described as short RP tachycardia, and the differentials would include typical AVNRT or atrioventricular re-entrant tachycardia (AVRT). If the RP interval is longer than the PR interval, the tachycardia is termed as long RP tachycardia, and the differentials would include atypical AVNRT or atrial tachycardia (AT) [1, 2]. The algorithm in **Figure 1** shows the flow chart of how to systematically approach a NCT.

**Figure 1.** Algorithm for narrow complex tachycardia.

## **2. Subtypes of NCTs**


#### **2.1. Sinus tachycardia**

hand, if the P waves are negative, that excludes a sinus origin of the rhythm. A negative P wave in front of the QRS complex implies a low atrial tachycardia (LAT). If the P waves embedded in the QRS complex have a short RP interval, it is described as short RP tachycardia, and the differentials would include typical AVNRT or atrioventricular re-entrant tachycardia (AVRT). If the RP interval is longer than the PR interval, the tachycardia is termed as long RP tachycardia, and the differentials would include atypical AVNRT or atrial tachycardia (AT) [1, 2]. The algorithm in **Figure 1** shows the flow chart of how to systemati-

cally approach a NCT.

4 Interpreting Cardiac Electrograms - From Skin to Endocardium

**2. Subtypes of NCTs**

**Figure 1.** Algorithm for narrow complex tachycardia.

**4.** Low atrial tachycardia

**5.** Typical AVNRT **6.** Atypical AVNRT **7.** Orthodromic AVRT **8.** Junctional tachycardia

**1.** Sinus tachycardia **2.** Atrial fibrillation **3.** Atrial flutter

Sinus tachycardia is a normal physiological response to exercise or pathology in which catecholamine release is enhanced. Sinus tachycardia is described as inappropriate sinus tachycardia (IST), when individuals present with chronic non-paroxysmal sinus tachycardia, with a structurally normal heart and with no apparent causes for sinus tachycardia. Sinus tachycardia is also noted in young individuals with structurally normal heart, in whom there is an exaggerated heart rate response to upright position, termed as postural orthostatic tachycardia syndrome (POTS). ECG in **Figure 2** is an example of a regular narrow complex tachycardia with upright P-waves in front of QRS complex, in leads I and II indicating a sinus origin with 1:1 relationship, consistent with sinus tachycardia. Vagal maneuvers or atrioventricular (AV) nodal blocking agents may be useful in further determining the mechanism of tachycardia. **Figure 3** is an ECG of a patient with cardiac tamponade, showing sinus tachycardia with electrical alternans.

#### **2.2. Atrial tachycardia (AT)**

AT is a regular atrial rhythm originating from an ectopic focus either from the right or left atrium. Focal AT is usually paroxysmal but can also present as incessant tachycardia, causing left ventricular systolic dysfunction. Right ATs are usually located in the tricuspid annulus followed by crista terminalis, coronary sinus ostium, and perinodal tissues in the order of frequency. Similarly, the left ATs predominantly originate from the pulmonary veins followed by mitral annulus, interatrial septum, and left atrial appendage. Classically, an ECG reveals negative P waves in the inferior leads, suggesting a caudo-cranial atrial activation (**Figure 4**).

**Figure 2.** ECG showing sinus tachycardia with upright P-wave before the QRS complex (blue arrow) indicating sinus origin with 1:1 AV relationship.

**Figure 3.** ECG showing sinus tachycardia with low QRS voltage and electrical alternans in a patient with cardiac tamponade.

#### **2.3. Atrioventricular nodal re-entrant tachycardia (AVNRT)**

Patients with AVNRT demonstrate dual AV nodal physiology, fast pathway with long refractory period and slow pathway with short refractory period. While 25–30% of the general population

**Figure 4.** ECG showing negative P-waves in the inferior leads and lead I with caudo-cranial activation consistent with low atrial tachycardia.

has dual AV nodal physiology, only a minority of them develop AVNRT. Typically, a critically timed premature atrial contraction (PAC) initiates the tachycardia and the classical symptom involves abrupt onset and offset of palpitations. Typical AVNRT accounts for 80–90% of all AVNRTs. In slow–fast AVNRT, the antegrade conduction occurs through the slow pathway and retrograde conduction through the fast pathway. The 12-lead ECG typically shows a narrow complex tachycardia with absent P waves or retrograde P waves and RP interval of <100 ms (**Figure 5**) [3]. In atypical AVNRT, the antegrade conduction occurs down the fast pathway and retrograde conduction up the slow pathway [4], and the 12-lead ECG will reveal a narrow complex tachycardia, with RP > PR interval, termed as long RP tachycardia.

#### **2.4. Atrioventricular re-entrant tachycardia (AVRT)**

**2.3. Atrioventricular nodal re-entrant tachycardia (AVNRT)**

6 Interpreting Cardiac Electrograms - From Skin to Endocardium

tamponade.

low atrial tachycardia.

Patients with AVNRT demonstrate dual AV nodal physiology, fast pathway with long refractory period and slow pathway with short refractory period. While 25–30% of the general population

**Figure 4.** ECG showing negative P-waves in the inferior leads and lead I with caudo-cranial activation consistent with

**Figure 3.** ECG showing sinus tachycardia with low QRS voltage and electrical alternans in a patient with cardiac

AVRT is a re-entrant tachycardia with a circuit that consists of two distinct pathways, normal AV nodal conduction system and the accessory pathway. The accessory pathway may manifest as preexcitation in the surface ECG (delta wave) if the antegrade conduction is exclusively through the pathway or a combination of pathway and AV node or concealed (without preexcitation) if the conduction is exclusively through the AV node. AVRT is classified as orthodromic if the antegrade conduction occurs via the AV node and antidromic if antegrade conduction occurs through the accessory pathway. Orthodromic AVRT accounts for 90–95% of the AVRTs associated with Wolf-Parkinson-White (WPW) syndrome [5]. **Figure 6** shows a routine surface ECG of a young patient performed prior to a dental extraction. An ECG reveals preexcitation with positive delta wave in inferior leads, I, aVL, and negative delta

**Figure 5.** ECG showing regular narrow complex tachycardia with retrograde P waves. Short RP (RP < PR) tachycardia with a RP interval of <100 ms.

**Figure 6.** Surface ECG showing preexcitation with positive delta wave in inferior leads and I, aVL and negative delta wave in V1 consistent with a right anteroseptal accessory pathway.

wave in V1 consistent with possible right anteroseptal accessory pathway. **Figures 7** and **8** show a narrow complex tachycardia with retrograde P waves and long ventriculoatrial (VA) time consistent with orthodromic AVRT.

**Figure 7.** SVT with longer VA time (>100 ms) consistent with orthodromic AVRT.

An Approach to Diagnosing Supraventricular Tachycardias on the 12-Lead ECG http://dx.doi.org/10.5772/intechopen.70143 9

**Figure 8.** Regular narrow complex tachycardia with retrograde P-waves (black arrows), RP < PR and RP > 100 ms.

## **3. SVT with aberrancy**

Narrow complex SVT may present as wide complex rhythm in the setting of aberrant conduction or the presence of accessory pathway. This can pose a diagnostic dilemma in differentiating from ventricular tachycardia (VT). Accurate differentiation of SVT-A from VT is challenging but is key to decision making and management in the emergency setting. In an effort to distinguish between VT and SVT with aberrancy, Brugada and Wellen proposed criteria to help determine the diagnosis [6].

Brugada and Wellen's criteria


**Figure 7.** SVT with longer VA time (>100 ms) consistent with orthodromic AVRT.

time consistent with orthodromic AVRT.

wave in V1 consistent with a right anteroseptal accessory pathway.

8 Interpreting Cardiac Electrograms - From Skin to Endocardium

wave in V1 consistent with possible right anteroseptal accessory pathway. **Figures 7** and **8** show a narrow complex tachycardia with retrograde P waves and long ventriculoatrial (VA)

**Figure 6.** Surface ECG showing preexcitation with positive delta wave in inferior leads and I, aVL and negative delta

An ECG in **Figure 9** is an example of SVT with aberrant conduction, on a patient who presented with palpitation. Rhythm strip shows regular P waves with 1:1 atrioventricular conduction. In addition, an ECG does not meet any of the above criteria for VT. An ECG in **Figure 10** shows a wide complex tachycardia with pseudo "RBBB" morphology, R to S >100 ms in the pseudo "LBBB" morphology leads, and 2:1 VA conduction, consistent with VT.

**Figure 9.** SVT with aberrancy.

**Figure 10.** ECG showing wide complex tachycardia, RBBB morphology, R to S >100 ms in V2, and 2:1 VA (blue arrows) conduction consistent with VT.

#### **4. Conclusion**

SVT includes a variety of tachycardias originating from the atria and atrioventricular nodal tissues. The subtypes of SVT differ in their physiological mechanism and presentation. The unpredictability of the episodes and inability to control the disabling symptoms can render the patient incapacitated. Accurate diagnosis of the subtypes of SVT is essential to deliver the most appropriate therapy. Despite the highly specified symptomatic and ECG diagnostic criteria, arriving at a diagnosis is not always straightforward. A systematic approach to interpreting ECG is therefore crucial to narrow the differential diagnosis. Careful interpretation of ECGs can unfold important information, depict the underlying mechanism, and aid with the diagnosis and management.

## **Author details**

Sanoj Chacko1 and Adrian Baranchuk2 \*

\*Address all correspondence to: barancha@kgh.kari.net

1 Heart Rhythm Service, Kingston General Hospital, Queen's University, Kingston, Ontario, Canada

2 Division of Cardiology, Department of Medicine, Kingston General Hospital, Queen's University, Kingston, Ontario, Canada

## **References**

**4. Conclusion**

conduction consistent with VT.

**Figure 9.** SVT with aberrancy.

10 Interpreting Cardiac Electrograms - From Skin to Endocardium

SVT includes a variety of tachycardias originating from the atria and atrioventricular nodal tissues. The subtypes of SVT differ in their physiological mechanism and presentation. The

**Figure 10.** ECG showing wide complex tachycardia, RBBB morphology, R to S >100 ms in V2, and 2:1 VA (blue arrows)


## **Chapter 2**

## **Examining Left Axis Deviation**

## Madhur Dev Bhattarai

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/intechopen.69435

#### **Abstract**

Axis deviation indicates possible presence of various conditions. It also affects the QRS and T morphologies. The limits of axis deviations are as such arbitrary and the approximate degree of axis itself can be easily determined. Various conditions often shift the QRS axis without fulfilling the defined limits of deviations in the initial stage. The associations with various conditions may be missed if such partial shift of the axis is disregarded. Isolated left axis deviation is relatively common in the general population and left anterior fascicular conduction delay is the most common cause of such isolated leftward shift of axis. Vulnerability of left anterior fascicle to interruption makes it likely to be affected by both atherosclerosis and fibrodegeneration. Glucose intolerance may increase the risk of both atherosclerosis and fibrodegeneration. The association of glucose intolerance with leftward shift of axis has been increasingly noticed. Research studies to get further evidences are required; however, utilizing the already available evidences to protect the susceptible population is equally essential. Documenting the approximate degree itself of the axis is the bottom line to study the association with the levels of various possible risk factors like glycated hemoglobin.

**Keywords:** ECG axis, left axis deviation, left anterior fascicular block, glucose intolerance, diabetes, diabetes prevention, indigenous population, ageing, Bachmann's bundle, neuropathy, white matter hyperintensities

## **1. Introduction**

The limits for mean frontal plane QRS axis deviations are considered variedly [1–11] and are necessarily arbitrary [5]. In this chapter, determination of left axis deviation and its effects on QRS‐T morphology and its causes will be discussed. The left axis deviation and leftward shift of axis have been increasingly noticed in asymptomatic relatively younger adults with diabetes. Next in the chapter, the epidemiology, pathogenesis, correlation with other related factors, and implications of the possible association between glucose intolerance and left axis deviation will be discussed.

© 2017 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

## **2. The conducting system of the heart**

In the heart, apart from ordinary myocardium and supporting fibrous skeleton, there are small groups of specialized neuromuscular cells in the myocardium which initiate and conduct cardiac electrical impulse [9, 12]. **Figure 1** shows the different parts of the specialized cardiac conducting system from the sinus node to the atrioventricular (AV) node with three internodal tracts in‐between and then from the AV node to the His‐Purkinje system [9, 12]. The atria and the ventricles are separated by a ring of fibrous tissue, which does not conduct electrical impulse. Thus, the electrical activity from the atria can only spread to the ventricles through the atrioventricular node and the atrioventricular bundle. Atrioventricular bundle (Bundle of His) originates from atrioventricular node and divides at the upper end of the ventricular system into right and left bundle branch [12]. The right

**Figure 1.** Different parts (shown in the boxes) of specialized cardiac conducting system from the sinoatrial node to the His‐Purkinje system.

\*Interatrial tract or the Bachmann's bundle is a branch of the anterior internodal tract to the left atrium which serves as the preferential path for electrical activation of the left atrium.

\*\*Fibrous ring does not conduct electrical impulse.

bundle branch does not divide, but the left bundle branch further divides into three separate fascicles, namely septal (median or medial) fascicle, left anterior fascicle, and left posterior fascicle. Right bundle branch and different fascicles of left bundle branch ultimately break up within the ventricular myocardium into fine fibers as the network of Purkinje fibers [12]. The left anterior fascicular block (or the left anterior hemiblock) causes left axis deviation and the left posterior fascicular block (or left posterior hemiblock) causes right axis deviation. Isolated left posterior fascicular block is extremely rare [2, 5]. Septal fascicle is found in nearly two‐thirds of people [13]. Delay or block of left septal fascicle may occur in diabetes, Chagas disease, and various cardiac diseases and may manifest in the ECG with prominent R waves in leads V1–V3, loss of septal Q waves, initial q waves in leads V1 and V2, and normal QRS axis [2]. However, the term left septal fascicular block is not recommended because of the lack of universally accepted criteria [3].

## **3. The normal mean frontal plane QRS axis range**

Sinoatrial (SA) node

(In the wall of right atrium near superior ventricular cava opening)

In the heart, apart from ordinary myocardium and supporting fibrous skeleton, there are small groups of specialized neuromuscular cells in the myocardium which initiate and conduct cardiac electrical impulse [9, 12]. **Figure 1** shows the different parts of the specialized cardiac conducting system from the sinus node to the atrioventricular (AV) node with three internodal tracts in‐between and then from the AV node to the His‐Purkinje system [9, 12]. The atria and the ventricles are separated by a ring of fibrous tissue, which does not conduct electrical impulse. Thus, the electrical activity from the atria can only spread to the ventricles through the atrioventricular node and the atrioventricular bundle. Atrioventricular bundle (Bundle of His) originates from atrioventricular node and divides at the upper end of the ventricular system into right and left bundle branch [12]. The right

> Atrioventricular (AV) node (In the wall of atrial septum near AV valve)

> > AV bundle (Bundle of His)

Right bundle branch Left bundle branch

(in 2/3rd of people)

Purkinje fibre network

**Figure 1.** Different parts (shown in the boxes) of specialized cardiac conducting system from the sinoatrial node to the

\*Interatrial tract or the Bachmann's bundle is a branch of the anterior internodal tract to the left atrium which serves as

Fibrous ring\*\* between atria and ventricles

> Left posterior fascicle

> Purkinje fibre network

Left anterior fascicle

Purkinje fibre network

Atria Anterior\*, middle, and posterior inter-nodal tracts

Fibrous ring\*\* between atria and ventricles

**2. The conducting system of the heart**

14 Interpreting Cardiac Electrograms - From Skin to Endocardium

Ventricles Septal fascicle

Purkinje fibre network

the preferential path for electrical activation of the left atrium.

\*\*Fibrous ring does not conduct electrical impulse.

His‐Purkinje system.

An electrocardiogram is a record of the origin and propagation of the electric action potential through heart muscle. Depolarization spreads throughout the heart to stimulate the myocardium to contract and the vector demonstrates the direction in which depolarization is moving. The general, mean, or dominant direction of all these vectors is known as the mean vector and is expressed electrocardiographically as the mean QRS axis [2] which is located by degrees [1]. The direction of the mean QRS axis on the frontal plane is known as the mean frontal plane QRS axis and is determined by the six frontal plane limb leads; they are three standard bipolar limb leads I, II, and III and three augmented unipolar limb leads aVR, aVL, and aVF. The frontal plane limb leads are conventionally represented on a hexaxial reference system (**Figure 2**).

In most normal adults, the mean QRS vector points downward and to the left [1] with the electric axis of the QRS complex almost parallel to the anatomic base‐to‐apex axis of the heart in the direction of the lead II [6]. Most normal frontal plane QRS axes in the adults are directed within a narrower range between +40° and +60° [2], around 5 o'clock position [7] (**Figure 2**). Such range has been reported particularly at sea level from studies of axes conducted at different altitudes [14, 15]. Leads I and aVF, II and aVL, and III and aVR are at right angles to each other; that is, each of the pair is the right‐angled partner leads. The concept of the right‐ angled partner leads (**Figure 2**) is helpful to quickly find the lead with the relatively tall R wave after looking at the lead with the equiphasic QRS complex. Coincidentally, but useful for remembering easily the pairs of the right‐angled partner leads, the letters F, L, and R of the augmented limb leads are in the increasing alphabetic order like the numbers I, II, and III of the bipolar limb leads.

There are variations not only in the conventionally considered limits of normal axis and left, right, and extreme axis deviations in the adults but also in the nomenclature of the deviations and in the use of positive and negative signs of the degrees of the axis. The indicated limits of normal axis and left, right, and extreme axis deviations in the adults by various publications are shown below; for example


The extreme axis deviation [4–6] has also been


Similarly, there are variations in the use of + and – signs. For 180°,


For other positive axis degrees many use + signs [2, 5–7, 9, 16] and some do not use + sign [3, 4].

In this chapter, *+ sign has been used for positive degree and – sign for negative degree*. As 0° and 180° are common to both positive and negative degree sides and there is no other similar degree to cause confusion, *no sign has been indicated for 0*° *and 180*°.

The consensus in these issues will help the communication among clinicians and between students and teachers especially during the examination of the student. However, even the consensually defined limits of normal axis and left, right, and extreme axis deviations should not make the clinicians and researchers to disregard the leftward or rightward shift of the axis from its usual range of degree (between +40° and +60°) in the adults (**Figure 2**). Otherwise the important correlation of the shift of axis with the patients' clinical condition or with the various factors in the research study may be missed; this is also further discussed later. With the possibility of easy determination of the approximate degree of axis by any clinicians and researches (see Section 4) and with the support of the computer interpretation of ECG readings at hand, the approximate degrees of the axis should be recorded.

**Figure 2.** The frontal plane limb leads conventionally represented on a hexaxial reference system showing the range of degrees (between +40° and +60°) of most normal frontal QRS axes in the adults (shown as the shaded area). Note: **The positive and negative poles of each lead:** The arrow head on the solid line designates the positive pole of the corresponding lead axis and the dotted line the negative pole. **The 30**° **differences:** There are 30° differences between the positive or negative poles of the nearby leads. **The right‐angled partner leads:** Leads I and aVF, II, and aVL, and III and aVR are at right angles to each other and the axis causing equiphasic deflections of QRS complex in one lead will cause maximum upward and downward deflections in the opposite ends of the other partner lead at right angle.

## **4. Determination of frontal plane QRS axis**

• **Normal axis** as 0° to +90° [1, 2], –30° to +90° [4, 6, 7, 10], and –30° to +100° [5, 9, 16].

• **Extreme axis deviation** as –90° to 180° [1, 2, 4–6] and –90° to –149° [10].

and –90° as *marked left axis deviations* [3].

16 Interpreting Cardiac Electrograms - From Skin to Endocardium

The extreme axis deviation [4–6] has also been

• still others use no + or – sign to 180° [4, 16].

• included under right axis deviation [7],

axis deviation [9], and

• many use ± [2, 6, 9],

[3, 4].

recorded.

• others use + sign [5], and

to 180° [9], +90° to –150° [10], and +90° to –90° [7].

• even indicated as the marked left or right axis deviation [5].

• some use both +180° and –180° in their axis range figure [7],

cause confusion, *no sign has been indicated for 0*° *and 180*°.

Similarly, there are variations in the use of + and – signs. For 180°,

• **Left axis deviation** as 0° to –90° [1, 2] and –30° to –90° [4–6, 10] with axis between 0° and –30° as *slight left axis deviation* [2], and between –30° and –45° as *moderate*, and between –45°

• **Right axis deviation** as +90° to 180° [2–4, 6], +90° to +150° [11], +100° to 180° [5, 16], +110°

• called as the northwest region axis [2], extreme right axis deviation [1], and extreme left

For other positive axis degrees many use + signs [2, 5–7, 9, 16] and some do not use + sign

In this chapter, *+ sign has been used for positive degree and – sign for negative degree*. As 0° and 180° are common to both positive and negative degree sides and there is no other similar degree to

The consensus in these issues will help the communication among clinicians and between students and teachers especially during the examination of the student. However, even the consensually defined limits of normal axis and left, right, and extreme axis deviations should not make the clinicians and researchers to disregard the leftward or rightward shift of the axis from its usual range of degree (between +40° and +60°) in the adults (**Figure 2**). Otherwise the important correlation of the shift of axis with the patients' clinical condition or with the various factors in the research study may be missed; this is also further discussed later. With the possibility of easy determination of the approximate degree of axis by any clinicians and researches (see Section 4) and with the support of the computer interpretation of ECG readings at hand, the approximate degrees of the axis should be

#### **4.1. Method of determining the approximate degree of mean axis**

When depolarization moves in a direction with the cardiac axis toward the positive pole of a lead, the deflection recorded by the lead is upward (positive), if it is away from the positive pole it is negative (downward) and if it is perpendicular to the orientation of a lead the deflection recorded by the lead is isoelectric or equiphasic (QR or RS) QRS with equal magnitudes of upward (positive) and negative (downward) deflection [5]. The leads between the one recording the equiphasic (QR or RS) deflection and the other recording the maximum upward deflection will record the increasing degree of upward deflection; this area can be designated as "*positive‐half area of the mean axis*" (**Figure 3**). Similarly, the leads

**Figure 3.** Varying degrees of upward or downward deflections recorded by different leads at various places in relation to the direction of cardiac axis (shown as the big central arrow). The leads in the "*positive‐half area of the mean axis*" and "*negative‐half area of the mean axis*" will record predominantly positive and negative QRS complex, respectively, increasingly so as per the distance from the points of equiphasic deflection (QR or RS).

between the one recording the equiphasic (QR or RS) deflection and the other recording the maximum downward deflection will record the increasing degree of downward deflection; this area can be designated as "*negative‐half area of the mean axis*." The mean QRS axis can, thus, be determined on the basis of one or both of the two rules [5]. As a general rule the mean QRS points


With these considerations, the mean frontal plane QRS axis can be determined with an error of 10–15° [5]. Thus, to determine the approximate degree of axis, one has to find the lead in which the QRS complex is most equiphasic (QR or RS); the axis is directed perpendicular to this lead (**Figure 4**). If the lead with clear equiphasic QRS complex is not seen, then the lead having the QRS with the largest positive amplitude should be looked for. If there are two nearby leads which have almost equal tall R wave, the axis is mostly directed in between them (**Figure 4**).

The degree of the mean axis can be further reconfirmed by considering whether it is in accord or not with the QRS direction and amplitude in other leads in the areas in front of the line of the equiphasic complex, that is, in the *positive‐half area of the mean axis* (**Figure 3**), and/or behind the line of the equiphasic complex, that is, in the *negative‐half area of the mean axis*. This

**Figure 4.** Method of determination of the approximate degree of the frontal plane QRS axis—The degree of the mean axis can be further reconfirmed by considering whether it is in accord or not with the QRS direction and amplitude in other leads in the areas in front of and/or behind the line of the equiphasic complex. \*For example, if the relatively tall R wave is in the lead aVL the axis is approximately –30°, if in I it is 0°, if in II it is +60°, if in aVF it is +90°, if in III it is +120°, and if in aVR it is –150°.

method is perhaps the most appropriate observational technique to follow in the routine setting. The other method of the mean axis determination by plotting the net height or depth of two standard bipolar (not the augmented) limb leads [8, 9] is generally not practiced.

Sometimes an electrocardiogram with indeterminate QRS axis is encountered. In indeterminate QRS axis, the algebraic sum of major positive and major negative QRS waves is zero in each of leads I, II, and III, or the information from these three leads is incongruous [10]. In the absence of a dominant QRS deflection, as in an equiphasic QRS complex, the axis is said to be indeterminate [9]. The separate determination of the axes of the initial and later part of QRS complex may indicate, or help to correlate with, other findings. For example, in right bundle branch block (RBBB) the axis determination is of importance in diagnosing associated left anterior or posterior fascicular block, as right bundle branch on its own will not cause axis deviation. In right bundle branch block, estimating the frontal plane QRS based on the first 80–100 ms of the QRS deflection, primarily reflecting activation of the left ventricle, may help [4]. For left bundle branch block (LBBB) and other intraventricular delays, the entire QRS or just the initial 80–100 ms can be used [4].

#### **4.2. An example of determination of the mean axis**

between the one recording the equiphasic (QR or RS) deflection and the other recording the maximum downward deflection will record the increasing degree of downward deflection; this area can be designated as "*negative‐half area of the mean axis*." The mean QRS axis can, thus, be determined on the basis of one or both of the two rules [5]. As a general rule the

**Figure 3.** Varying degrees of upward or downward deflections recorded by different leads at various places in relation to the direction of cardiac axis (shown as the big central arrow). The leads in the "*positive‐half area of the mean axis*" and "*negative‐half area of the mean axis*" will record predominantly positive and negative QRS complex, respectively,

• midway between the axes of two extremity leads that show tall R waves of equal ampli-

• at 90° (right angle) to any extremity lead that shows a biphasic (QR or RS) complex and in

With these considerations, the mean frontal plane QRS axis can be determined with an error of 10–15° [5]. Thus, to determine the approximate degree of axis, one has to find the lead in which the QRS complex is most equiphasic (QR or RS); the axis is directed perpendicular to this lead (**Figure 4**). If the lead with clear equiphasic QRS complex is not seen, then the lead having the QRS with the largest positive amplitude should be looked for. If there are two nearby leads which have almost equal tall R wave, the axis is mostly directed in between them

The degree of the mean axis can be further reconfirmed by considering whether it is in accord or not with the QRS direction and amplitude in other leads in the areas in front of the line of the equiphasic complex, that is, in the *positive‐half area of the mean axis* (**Figure 3**), and/or behind the line of the equiphasic complex, that is, in the *negative‐half area of the mean axis*. This

the direction of leads that show relatively tall R waves [5].

increasingly so as per the distance from the points of equiphasic deflection (QR or RS).

18 Interpreting Cardiac Electrograms - From Skin to Endocardium

mean QRS points

tude, and

(**Figure 4**).

An ECG with the recordings of the limb leads in an asymptomatic person is shown in **Figure 5** with left axis deviation. The QRS complex in II (which is at +60° in the hexaxial reference system) is most equiphasic; thus, the axis will be 90° to it, either in –30° or +150° (**Figure 2**). Since R wave in the lead aVL is positive and has almost the largest R wave amplitude, it dictates the direction of the vector. Thus, the QRS axis is leftward between around –30°. However, if we look carefully the lead II is not exactly equiphasic but it has slightly more negative complex especially the second QRS complex, so the lead II lies in the "*negative‐half area of the mean axis*" (**Figure 3**). Similarly, the lead aVR is also near equiphasic with slightly more negative QRS complex, that is, the lead aVR is also in the "*negative‐half area of the mean axis*" (**Figure 3**). As the

**Figure 5.** An electrocardiogram recording of the limb leads in an asymptomatic person.

lead II has slightly less negative QRS complex than the lead aVR, the lead II is slightly nearer to the mean axis than the aVR. So the mean axis is most likely around –40°.

In this example, we have determined the mean axis by first looking at the lead with the equiphasic QRS complex as described in **Figure 4** and later also reconfirm the degree of the mean axis by considering whether the QRS direction and amplitude in the leads in the *negative‐half area* of the mean axis are in accord or not with it. As discussed above, the approximate determination of the axis with such method entails an error of 10–15° [5]. Moreover, minor degrees of change in the height of QRS complex may also occur due to the difference in the relative voltage and magnitude of the bipolar and augmented unipolar leads [9] or due to cardiac or respiratory movement.

#### **4.3. Other axes**

When cardiac axis is referred, it usually indicates the mean frontal plane QRS axis. But there are also other axes to consider, for example, T wave and P wave axes. Though even axis of ST segment [2] and the QRS axis in the horizontal plane [1, 6, 9] are also discussed, such axes are not used in the routine ECG interpretation. The axes of T wave and P wave are also not routinely determined. However, in the computer interpretation in modern electrocardiogram tracings the axes of P and T waves are now easily available. Thus, routine consideration of T and P wave axes is possible and may be helpful. T wave axis is discussed later in Section 5.3. The normal mean manifest *frontal plane P wave axis is also directed generally within the region of +40*° *to +60*° [2] *with the normal limits between 0*° *and +75*° [6]. The axis of P wave is affected by different conditions [2, 6], for example,


#### **4.4. Usefulness and limitations of the computer interpretations in modern electrocardiograms**

The computer interpretations in modern electrocardiogram tracings help to note if anything is missed or to crosscheck the findings. But that does not decrease the responsibility of the physicians. In the textbook on electrocardiogram prepared on behalf of the council on clinical cardiology of the American Heart Association, it is emphasized: "*Even though computer interpretations of ECGs are readily available, the clinician's role as overseer and final interpreter has not and must not be diminished*" [9]. On the other hand, one should also not totally ignore the computer interpretations. The computer interpretations are particularly useful for *rate*, *axes of QRS*, *P and T waves*, *intervals and amplitude*, *or voltage and duration of different waves*. Though in the computer interpretation the range of normal values is not given, the possible abnormal findings are indicated. The overall computer interpretation in the modern electrocardiogram machines, however, cannot incorporate the various clinical conditions to be considered in each individual patient. The rhythm, P, QRS, and T wave morphology, and ST segment need to be given particular attention by the clinicians and the conclusion should be drawn considering all the clinical conditions of the individual patient.

## **5. Effect of axis deviation on QRS‐T morphology**

lead II has slightly less negative QRS complex than the lead aVR, the lead II is slightly nearer

In this example, we have determined the mean axis by first looking at the lead with the equiphasic QRS complex as described in **Figure 4** and later also reconfirm the degree of the mean axis by considering whether the QRS direction and amplitude in the leads in the *negative‐half area* of the mean axis are in accord or not with it. As discussed above, the approximate determination of the axis with such method entails an error of 10–15° [5]. Moreover, minor degrees of change in the height of QRS complex may also occur due to the difference in the relative voltage and magnitude of the bipolar and augmented unipolar leads [9] or due to cardiac or

When cardiac axis is referred, it usually indicates the mean frontal plane QRS axis. But there are also other axes to consider, for example, T wave and P wave axes. Though even axis of ST segment [2] and the QRS axis in the horizontal plane [1, 6, 9] are also discussed, such axes are not used in the routine ECG interpretation. The axes of T wave and P wave are also not routinely determined. However, in the computer interpretation in modern electrocardiogram tracings the axes of P and T waves are now easily available. Thus, routine consideration of T and P wave axes is possible and may be helpful. T wave axis is discussed later in Section 5.3. The normal mean manifest *frontal plane P wave axis is also directed generally within the region of +40*° *to +60*° [2] *with the normal limits between 0*° *and +75*° [6]. The axis of P wave is affected by

• +60° to +90° or even more to right axis deviation in acquired right heart diseases including due to chronic obstructive pulmonary disease (COPD) with tall upright P wave in II, aVF,

• –80° to –90° in the retrograde activation of the atria by an impulse from AV node or below

and III leads with near equiphasic or equiphasic P waves in lead I, • +45° to –30° in left atrial diseases with upright in I and aVL leads, and

to the mean axis than the aVR. So the mean axis is most likely around –40°.

**Figure 5.** An electrocardiogram recording of the limb leads in an asymptomatic person.

20 Interpreting Cardiac Electrograms - From Skin to Endocardium

respiratory movement.

different conditions [2, 6], for example,

with inverted P wave in II, aVF, and III leads.

**4.3. Other axes**

#### **5.1. Variations in the QRS pattern and usually negative waves in aVR**

The QRS pattern in the extremity leads may vary considerably from one normal subject to another depending on the electrical axis of the QRS, which describes the mean orientation of the QRS vector with reference to the six frontal plane leads [16]. The effects on QRS and T morphologies by different frontal plane QRS axes may thus cause confusion in the interpretation of ECG. A shift, even a pronounced one, of the heart to the right due to pneumothorax or pleural effusion, however, does not necessarily affect the frontal plane QRS axis [17]. The effect of axis in the QRS and T morphologies can also be utilized for the efficient interpretation of electrocardiogram. The frontal plane QRS axis in most people are directed away from aVR, thus in the lead aVR the QRS complex is mostly negative. Marked left axis (<–60°) or right axis (>+120°) or extreme axis deviation will cause positive QRS complex in aVR. And the dominantly upright QRS as well as P waves in aVR along with the inverted ones in leads I and aVL indicate the presence of reversal of the left and right arm electrodes or rarely the dextrocardia, a common form of cardiac malposition [2, 18]. Thus, while interpreting any ECG it is helpful to look first at aVR to quickly detect the incorrectly placed arm electrodes or the marked left or right axis deviation.

#### **5.2. An example of the effect on QRS complex due to left axis deviation**

The ECG with the recordings of the limb leads shown in **Figure 5** is used as an example to discuss the effect of left axis deviation on the appearance of QRS complex. In the ECG shown in **Figure 5**, the QRS complexes in III and aVF appear mostly negative raising the possibility of inferior wall ischemia. But this pattern in the asymptomatic person is most likely due to the frontal plane mean axis being at around –40°. The equiphasic points for this mean axis of –40° are at right angle, that is, at +50° and at –130°. The leads aVF and III are away from the equiphasic point at +50° toward the opposite pole of mean axis in the "*negative‐half area of the mean axis*" (**Figure 3**) and are recording the increasingly negative QRS complex. Thus, the QRS complexes in III and aVF appear mostly negative. An ECG with right axis deviation is similarly likely to cause predominantly negative QRS complex in the leads I and aVL.

#### **5.3. An example of the effect on T wave due to left axis deviation**

Similarly the T waves in leads aVF and III in the ECG in **Figure 3** appear flat or inverted raising the possibility of inferior wall ischemia. However, as a rule the mean T wave axis and the mean QRS axis normally point in the same general (but not identical) direction [5]. Thus, in general when the main QRS deflection is positive (upright), the T wave is normally also positive [5] and vice versa. *In the adult, the QRS‐T angle normally does not exceed 45*° *in the frontal plane [6]. When the angle is greater than 60*°*, the electrocardiogram is abnormal and disease is usually present [2].* In the ECG in **Figure 5**, the mean axis is directed at around –40° with the QRS complexes in III and aVF appearing more negative. In the asymptomatic person with this ECG, the mean frontal plane T wave axis is likely to be directed as the mean frontal plane QRS axis with the T wave appearing flat or inverted like the QRS complex. The T wave axis given in the computer interpretation in the modern electrocardiogram may provide additional information to correlate with the clinical condition of the patient.

#### **5.4. The variations in the QRS‐T pattern and its implications**

The mean axis of the QRS complex causing the variations in the QRS‐T pattern is just one of the many factors to do so. Minor degrees of or apparent ST segment or T wave changes can also occur due to many general medical or cardiac conditions other than the ischemic ones. Various such conditions related to each individual patient are not incorporated in the computer interpretation of the ECG tracing. The clinician has to correlate the ECG findings with the clinical conditions of the patient.

The US Preventive Services Task Force (USPTF) recommends against screening with resting or exercise ECG for the prediction of coronary heart disease events in asymptomatic adults at low risk for coronary heart disease events. The USPSTF concludes that the current evidence is insufficient to assess the balance of benefits and harms of screening with resting or exercise ECG for the prediction of coronary heart disease events in asymptomatic adults at intermediate or high risk for coronary heart disease events [19]. The overall risk for a serious adverse event, one that requires hospitalization or causes sudden death, with exercise ECG is estimated to be up to 1 in 10,000 tests [20].

Moreover, up to 71% of asymptomatic adults with abnormal exercise treadmill ECG results have no angiographically demonstrable coronary artery stenosis [21] and revascularization as such has also not been shown to reduce coronary heart disease events in asymptomatic persons [19]. And the risk for any serious adverse event from angiography is up to 1.7%, including risk for death (0.1%), myocardial infarction (0.05%), stroke (0.07%), or arrhythmia (0.4%) [19, 22]. Such complications are more likely in the countries where the skills of angiography and angioplasty are not certified with appropriate criteria (considering *the learning curve* required for the particular skill) and in the situation where *the volume of the procedures* performed by the operator or in the institutions are not sufficient.

## **6. Causes of left axis deviation**

The conditions which are likely to fulfill the criteria of the defined cut‐off points and are conventionally considered as the causes of left axis deviation [1, 2, 4–8, 17] are shown in **Table 1**. An individual patient may have more than one medical condition; in such conditions, the resultant axis deviation may not reflect the typical pattern of one cause.

Axis deviation in conditions like right bundle branch block (RBBB) and left bundle branch block (LBBB) also depends on their underlying causes. LBBB is commonly a sign of organic heart disease [2] and most patients with LBBB have underlying left ventricular hypertrophy [5]. Thus, left axis deviation is usually seen in LBBB. Though LBBB is even emphasized to be always a sign of heart disease usually of the left ventricle [7], it is rarely seen in normal individuals without any organic heart disease [5]. Some other congenital heart diseases where left axis deviation is seen include

• tricuspid atresia,

of inferior wall ischemia. But this pattern in the asymptomatic person is most likely due to the frontal plane mean axis being at around –40°. The equiphasic points for this mean axis of –40° are at right angle, that is, at +50° and at –130°. The leads aVF and III are away from the equiphasic point at +50° toward the opposite pole of mean axis in the "*negative‐half area of the mean axis*" (**Figure 3**) and are recording the increasingly negative QRS complex. Thus, the QRS complexes in III and aVF appear mostly negative. An ECG with right axis deviation is simi-

Similarly the T waves in leads aVF and III in the ECG in **Figure 3** appear flat or inverted raising the possibility of inferior wall ischemia. However, as a rule the mean T wave axis and the mean QRS axis normally point in the same general (but not identical) direction [5]. Thus, in general when the main QRS deflection is positive (upright), the T wave is normally also positive [5] and vice versa. *In the adult, the QRS‐T angle normally does not exceed 45*° *in the frontal plane [6]. When the angle is greater than 60*°*, the electrocardiogram is abnormal and disease is usually present [2].* In the ECG in **Figure 5**, the mean axis is directed at around –40° with the QRS complexes in III and aVF appearing more negative. In the asymptomatic person with this ECG, the mean frontal plane T wave axis is likely to be directed as the mean frontal plane QRS axis with the T wave appearing flat or inverted like the QRS complex. The T wave axis given in the computer interpretation in the modern electrocardiogram may provide additional informa-

The mean axis of the QRS complex causing the variations in the QRS‐T pattern is just one of the many factors to do so. Minor degrees of or apparent ST segment or T wave changes can also occur due to many general medical or cardiac conditions other than the ischemic ones. Various such conditions related to each individual patient are not incorporated in the computer interpretation of the ECG tracing. The clinician has to correlate the ECG findings with

The US Preventive Services Task Force (USPTF) recommends against screening with resting or exercise ECG for the prediction of coronary heart disease events in asymptomatic adults at low risk for coronary heart disease events. The USPSTF concludes that the current evidence is insufficient to assess the balance of benefits and harms of screening with resting or exercise ECG for the prediction of coronary heart disease events in asymptomatic adults at intermediate or high risk for coronary heart disease events [19]. The overall risk for a serious adverse event, one that requires hospitalization or causes sudden death, with exercise ECG is esti-

Moreover, up to 71% of asymptomatic adults with abnormal exercise treadmill ECG results have no angiographically demonstrable coronary artery stenosis [21] and revascularization as such has also not been shown to reduce coronary heart disease events in asymptomatic persons [19]. And the risk for any serious adverse event from angiography is up to 1.7%, including risk for death (0.1%), myocardial infarction (0.05%), stroke (0.07%), or arrhythmia

larly likely to cause predominantly negative QRS complex in the leads I and aVL.

**5.3. An example of the effect on T wave due to left axis deviation**

22 Interpreting Cardiac Electrograms - From Skin to Endocardium

tion to correlate with the clinical condition of the patient.

the clinical conditions of the patient.

mated to be up to 1 in 10,000 tests [20].

**5.4. The variations in the QRS‐T pattern and its implications**


**Table 1.** Causes of left axis deviation.

In chronic obstructive pulmonary disease (COPD) with development of pulmonary hypertension, the frontal plane QRS axis shifts to the right side, even to +150° [2]. In very severe cases, extreme axis deviation may occur. Occasionally, in about 10% cases, there may be left axis deviation, the frontal plane QRS axis being directed upwards and to the left in the vicinity of –60° to –90° [2]. The exact mechanism is uncertain, but it is also supposed to be *an axis illusion* and the term *pseudo‐left axis deviation* has also been applied [17]. In an individual patient with COPD, the presence of left axis deviation rather points out the need to look for other conditions, especially coronary heart disease, due to the common etiological factor of smoking shared by both diseases.

## **7. The leftward shift of axis**

While enumerating the causes of axis deviation there may be confusion to


There are different physiological and pathological conditions which may shift the axis to one side or the other.

#### **7.1. Effect of breathing**

When a person breathes in, the diaphragm descends and the heart becomes more vertical and QRS electrical axis generally shifts toward right side, and when the person breathes out, the diaphragm ascends, heart assumes a more horizontal position, and the axis shifts toward the left side [5]. This may be more pronounced in maximum inspiration and expiration.

#### **7.2. Body habitus and obesity**

The mean frontal plane electrical axis also depends on body habitus [3]. The axes are more vertical in thin individuals and more horizontal in heavy individuals [6]. Obesity may deviate the frontal plane QRS axis toward the left side but not further to the left than 0° and a left axis deviation to –30° or further leftwards in an obese person probably represents a pathological abnormality [2, 23].

#### **7.3. Pregnancy**

A small rightward QRS axis shift may occur in the first trimester [24]. Similarly, some degree of leftward shift of QRS axis by about 15–30° has been reported in the third trimester of pregnancy by different studies [24–26] and the axis shifting back to normal side after delivery [26]. However, in some cases of pregnancy, a slight rightward shift of QRS axis (within the normal range) may occur at full term [27]. It is emphasized that pregnancy is not associated with left axis deviation or any significant change of QRS axis [2]. Thus, in general some degree of leftward shift of QRS axis may occur in the third trimester of pregnancy. However, an isolated rightward QRS axis change may be encountered in normal pregnant patients and cannot be viewed as a definite abnormality or used as a sole criterion for heart disease [24].

#### **7.4. The causes of left axis deviation**

In chronic obstructive pulmonary disease (COPD) with development of pulmonary hypertension, the frontal plane QRS axis shifts to the right side, even to +150° [2]. In very severe cases, extreme axis deviation may occur. Occasionally, in about 10% cases, there may be left axis deviation, the frontal plane QRS axis being directed upwards and to the left in the vicinity of –60° to –90° [2]. The exact mechanism is uncertain, but it is also supposed to be *an axis illusion* and the term *pseudo‐left axis deviation* has also been applied [17]. In an individual patient with COPD, the presence of left axis deviation rather points out the need to look for other conditions, especially coronary heart disease, due to the common etiological factor of smoking shared by both

• whether the conditions mentioned are just likely to shift the axis toward one or other side of the usual normal axis between +40° and +60° (**Figure 2**) but within the conventionally

• whether the conditions mentioned shift the axis frankly toward the right or left side fulfill-

There are different physiological and pathological conditions which may shift the axis to one

When a person breathes in, the diaphragm descends and the heart becomes more vertical and QRS electrical axis generally shifts toward right side, and when the person breathes out, the diaphragm ascends, heart assumes a more horizontal position, and the axis shifts toward the

The mean frontal plane electrical axis also depends on body habitus [3]. The axes are more vertical in thin individuals and more horizontal in heavy individuals [6]. Obesity may deviate the frontal plane QRS axis toward the left side but not further to the left than 0° and a left axis deviation to –30° or further leftwards in an obese person probably represents a pathological

A small rightward QRS axis shift may occur in the first trimester [24]. Similarly, some degree of leftward shift of QRS axis by about 15–30° has been reported in the third trimester of pregnancy by different studies [24–26] and the axis shifting back to normal side after delivery [26].

left side [5]. This may be more pronounced in maximum inspiration and expiration.

While enumerating the causes of axis deviation there may be confusion to

diseases.

**7. The leftward shift of axis**

24 Interpreting Cardiac Electrograms - From Skin to Endocardium

ing the defined limits of the axis deviations.

considered limits or

side or the other.

**7.1. Effect of breathing**

**7.2. Body habitus and obesity**

abnormality [2, 23].

**7.3. Pregnancy**

The conditions which are likely to fulfill the criteria of the defined cut‐off points and are conventionally considered as the causes of left axis deviation [1, 2, 4–8, 17] (**Table 1**) may also not fulfill the defined criteria especially in the initial stages but they may shift the axis to one side from its usual range of the degrees between +40° and +60° (**Figure 2**). Most causes of left axis deviation (**Table 1**) are well‐known clinical entities. Isolated left axis deviation and leftward shift of axis have been increasingly noticed by the author especially in relatively young adults with diabetes. There are many points to consider regarding leftward shift of axis and particularly the isolated left anterior fascicular block.

## **8. The isolated left anterior fascicular conduction delay**

There are different causes of left axis deviation (**Table 1**). Left axis deviation may also occur in the absence of apparent cardiac disease and it is not necessarily a sign of significant underlying heart disease [5]. Left axis deviation is relatively common with advancing age even in the absence of clinically overt heart disease and rare during early adult years [17, 28–40]. In a population‐based study of the people 20 years and above, almost half of the people with left axis deviation had isolated left axis deviation without evidence of heart diseases [28]. Left anterior fascicular block is the most common cause of left axis deviation [2]. The classical criteria of left anterior fascicular block are frontal plane axis between –45° and –90°, qR pattern in lead aVL, R peak time in lead aVL of 45 ms or more, and QRS duration less than 120 ms [3]. There are different causes of left anterior fascicular block or conduction delay [2, 5, 17, 27] (**Table 2**). In neuromuscular diseases like myotonia dystrophica, the involvement appears to be in the cardiac conduction system as a sort of nonmyopathic manifestations [17]. Hyperkalemia as well as a sudden increase in serum potassium levels are sometimes accompanied by left axis deviation ascribed to left anterior fascicular block due to changes in resting membrane potential and transmembrane potassium gradient; similar mechanism is held responsible for the generalized QRS widening with hyperkalemia [17].

The finding of isolated left anterior fascicular block is a very common, nonspecific abnormality [5]. The QRS axes which range from 0° to –30° probably reflect minor degrees of left anterior fascicular block or incomplete left anterior fascicular block [2]. An axis of +29° is also considered as already reflecting some degree of left axis deviation [2]. Complete block of conduction in the left anterior fascicle is not necessary to produce left axis deviation; presumably, all that is required


**Table 2.** Causes of left anterior fascicular block or conduction delay.

is enough delay in anterior fascicular conduction to result in the activation of the anterior left ventricular solely via the posterior fascicle [17]. In fascicular block, left axis deviation can be interpreted as either delayed conduction or complete block in the left anterior fascicle [17], which may explain the leftward shift of axis from minor degrees to frank left axis deviation.

## **9. Vulnerability of left anterior fascicle**

The left anterior fascicle is more vulnerable to interruption than the left posterior fascicle because of many reasons [2, 17] (**Table 3**). The left anterior fascicle is often supplied by septal branch of descending artery only or by septal branch of descending artery and atrioventricular node artery or rarely by atrioventricular node artery alone [17]. A total occlusion of the left anterior descending artery may cause a subsequent right bundle branch block with left anterior fascicular block [1]. This is one of the reasons for the frequent manifestation of right bundle branch block with left anterior fascicular block [2].


**Table 3.** Reasons of left anterior fascicular being more vulnerable to interruption than the left posterior fascicle.

## **10. Isolated left anterior fascicular conduction delay due to atherosclerosis and degenerative conditions**

The causes and clinical significance of left axis deviation have always been of interest since the early days of electrocardiography [17]. In regards to the causation of isolated left anterior fascicular delayed conduction or block, the possible mechanisms are fibrosis related to atherosclerosis and degenerative conditions [17]. The vulnerability of left anterior fascicle to interruption (**Table 3**) indicates the possibility of atherosclerosis with resultant coronary heart diseases. The possibility has, thus, been pointed out that left anterior fascicular block which occurs in the elderly may be due to subclinical coronary artery disease [2]. Ischemic heart disease in its own right causes fibrosis that partially or completely interrupts conduction in one or more fascicles [17]. On the other hand, fibrosis and degenerative disorder of the anterior fascicle of left bundle branch are postulated to be the cause of left axis deviation in the older population without associated cardiovascular abnormalities [17, 37–41]. The ECG trend of the gradual leftward migration of the frontal QRS axis has been concluded to be a common sequel of aging, independent of the population prevalence of coronary atherosclerosis. Thus, isolated, age‐related degenerative disease is also considered to cause a variety of infranodal conduction defects that are unrelated to coexisting myocardial disease or coronary artery obstruction, which may be negligible or absent [17].

**Interatrial conduction block by fibrosis**: It is interesting to note here that the association between conduction delays and block in Bachmann's bundle (**Figure 5**) and atrial fibrillation has been reported [42]. The Bachmann's bundle is recognized as a muscular bundle and shares electrophysiological properties of both Purkinje and atrial fibers; it is not surrounded by a fibrous tissue sheath. Fibrosis of the interatrial tract or Bachmann's bundle has been suggested as the mechanism underlying interatrial conduction block. Areas of conduction block may not be confined to Bachmann's bundle alone [42]. The association of such changes with the left axis deviation and left anterior fascicular conduction delay deserves study.

## **11. Left axis deviation and glucose intolerance**

is enough delay in anterior fascicular conduction to result in the activation of the anterior left ventricular solely via the posterior fascicle [17]. In fascicular block, left axis deviation can be interpreted as either delayed conduction or complete block in the left anterior fascicle [17], which may explain the leftward shift of axis from minor degrees to frank left axis deviation.

• Ageing

• Atherosclerosis

duction system

• Rarely in hyperkalemia

• Long‐standing hypertension

Lenegre disease or Lev disease

• Other secondary degenerative disorders of the con-

• Primary sclerodegenerative disorders, for example,

• Congenital isolated left anterior fascicular block • Neuromuscular diseases like myotonia dystrophica, peroneal muscular atrophy, limb‐girdle dystrophy

The left anterior fascicle is more vulnerable to interruption than the left posterior fascicle because of many reasons [2, 17] (**Table 3**). The left anterior fascicle is often supplied by septal branch of descending artery only or by septal branch of descending artery and atrioventricular node artery or rarely by atrioventricular node artery alone [17]. A total occlusion of the left anterior descending artery may cause a subsequent right bundle branch block with left anterior fascicular block [1]. This is one of the reasons for the frequent manifestation of right

Length and width Relatively long and thin Relatively short and thick Blood supply Mostly single blood supply (see text) Dual blood supply

aortic valve disease and surgery)\*

**Table 3.** Reasons of left anterior fascicular being more vulnerable to interruption than the left posterior fascicle.

**Left anterior fascicle Left posterior fascicle**

Further away from aortic valve\*

**9. Vulnerability of left anterior fascicle**

**Table 2.** Causes of left anterior fascicular block or conduction delay.

**Clinical cardiac diseases Other conditions**

• Coronary heart disease • Left ventricular hypertrophy • Chronic cardiac failure • Cardiomyopathies • Aortic valve diseases

• Infiltrative diseases • Focal pathological lesions

• Various congenital heart diseases

• Cardiac surgery (especially aortic valve surgery)

26 Interpreting Cardiac Electrograms - From Skin to Endocardium

bundle branch block with left anterior fascicular block [2].

Proximity to aortic valve Closer (so likely to be affected by

\*Left anterior fascicle is situated superiorly and left posterior fascicle inferiorly.

In a study of almost the entire population aged 16 years and above in a town in 1959–1960 among people with left axis deviation, more than 25%, and among those less than 40 years of age, 36–40%, have hyperglycemia with blood glucose value above the 80th percentile for the age group [32]. Similarly, in another study of people with diabetes and control group, diabetic men have more leftward frontal QRS axis than their nondiabetic counterparts when the effect of confounding factors (age, obesity, coronary heart disease, hypertension, and drugs) was taken into account [43]. In a population study, among people with isolated left axis deviation almost half (47.4%) of the persons less than 40 years age have blood glucose in the upper quintile values in comparison to 20.7% of those more than 50 years [28]. In a study of asymptomatic people aged 30 years or more not on any medication attending outdoor clinics for health checkup, the mean (SD) values of fasting plasma glucose are 101.0 ± 18.3 mg/dL in the slight left axis deviation group with QRS axis 0° to –30° (mean age 40.3 ± 8.5) and 122.9 ± 27.5 mg/dL in moderate‐to‐marked left axis deviation group with QRS axis –30° to –90° (mean age 54.5 ± 6.3). The frequency of glucose intolerance is 48.9% in the slight left axis deviation group with QRS axis 0° to –30° and 84.9% in moderate‐to‐marked left axis deviation group with QRS axis –30° to –90°, the difference being significant after conditioning the effects of age and sex (*P* ≤ 0.03) and after conditioning the effect of BP (*P* = 0.02) [44, 45]. In a recent study with 85% of participants less than 55 years of age, left axis deviation was present in 8% of control group and 43.3% of type 2 diabetes [46]. The frequency of left axis deviation in the control nondiabetic group mostly below the age of 55 years in this report is also relatively high. The control group, though do not have diabetes, may have higher level of glucose or glycated hemoglobin (HbA1c) which could be related to the increasing glucose intolerance in the population now.

## **12. Why glucose intolerance as a cause of left anterior fascicular conduction delay was not much reported in earlier reports?**

#### **12.1. Difficulty in conducting fasting and 2‐hour glucose estimation for diagnosis in the studies**

In many reports of different findings in ECG of varied populations, plasma glucose estimation was mostly not done [29, 31, 33–40]. Conducting fasting and 2‐hour glucose estimation in the field situation may not be easy as people have to come in the fasting state and wait for further 2 hours after taking glucose. Glycated hemoglobin (HbA1c) has been recently recommended for diagnosis of glucose intolerance.

#### **12.2. A relatively new phenomenon**

The epidemic of glucose intolerance in the world is relatively a new phenomenon starting since the latter half of twentieth century [47, 48]. It is now increasingly affecting the younger population [47]. The situation of the epidemic of diabetes could have led to observe and report the association of glucose intolerance and left axis deviation in relatively younger people more explicitly now [45, 46].

#### **12.3. Data collection only as the normal axis or as the left axis deviation mostly with –30° to –90°**

In most studies, including that of people with hyperglycemia or diabetes, only the presence of left axis deviation with QRS frontal plane axis –30° to –90° is considered [29–34, 36–39, 49], not the leftward shift of axis from its usual normal position (**Figure 2**), and thus the lower range of leftward shift is likely to be missed. The process of gradual shifting of the ECG axis toward left could be associated with the period of exposure to different grades and combination of the related factors like increasing age, glucose level, and other factors. Focusing only on the left axis deviation criteria as –30° to –90° by the studies also appears as one reason of dearth of evidences about the association of different degrees of QRS axis with possible factors. The arbitrary limits of the axis deviations have already been discussed.

## **13. Left axis deviation in school children in indigenous population**

outdoor clinics for health checkup, the mean (SD) values of fasting plasma glucose are 101.0 ± 18.3 mg/dL in the slight left axis deviation group with QRS axis 0° to –30° (mean age 40.3 ± 8.5) and 122.9 ± 27.5 mg/dL in moderate‐to‐marked left axis deviation group with QRS axis –30° to –90° (mean age 54.5 ± 6.3). The frequency of glucose intolerance is 48.9% in the slight left axis deviation group with QRS axis 0° to –30° and 84.9% in moderate‐to‐marked left axis deviation group with QRS axis –30° to –90°, the difference being significant after conditioning the effects of age and sex (*P* ≤ 0.03) and after conditioning the effect of BP (*P* = 0.02) [44, 45]. In a recent study with 85% of participants less than 55 years of age, left axis deviation was present in 8% of control group and 43.3% of type 2 diabetes [46]. The frequency of left axis deviation in the control nondiabetic group mostly below the age of 55 years in this report is also relatively high. The control group, though do not have diabetes, may have higher level of glucose or glycated hemoglobin (HbA1c) which could be related

to the increasing glucose intolerance in the population now.

mended for diagnosis of glucose intolerance.

28 Interpreting Cardiac Electrograms - From Skin to Endocardium

**12.2. A relatively new phenomenon**

more explicitly now [45, 46].

**to –90°**

**in the studies**

**12. Why glucose intolerance as a cause of left anterior fascicular** 

**12.1. Difficulty in conducting fasting and 2‐hour glucose estimation for diagnosis** 

In many reports of different findings in ECG of varied populations, plasma glucose estimation was mostly not done [29, 31, 33–40]. Conducting fasting and 2‐hour glucose estimation in the field situation may not be easy as people have to come in the fasting state and wait for further 2 hours after taking glucose. Glycated hemoglobin (HbA1c) has been recently recom-

The epidemic of glucose intolerance in the world is relatively a new phenomenon starting since the latter half of twentieth century [47, 48]. It is now increasingly affecting the younger population [47]. The situation of the epidemic of diabetes could have led to observe and report the association of glucose intolerance and left axis deviation in relatively younger people

**12.3. Data collection only as the normal axis or as the left axis deviation mostly with –30°** 

In most studies, including that of people with hyperglycemia or diabetes, only the presence of left axis deviation with QRS frontal plane axis –30° to –90° is considered [29–34, 36–39, 49], not the leftward shift of axis from its usual normal position (**Figure 2**), and thus the lower range of leftward shift is likely to be missed. The process of gradual shifting of the ECG axis toward left could be associated with the period of exposure to different grades and combination of

**conduction delay was not much reported in earlier reports?**

High prevalence of left axis deviation, 6–9‐fold higher than the control group, in healthy American‐Indian Navajo and Apache school children has been reported, the possible cause of which was considered unexplained [50]. Mean frontal plane QRS axis between –1° and –90° was present in 19% of the Navajo and 12% of the Apache school children. The prevalence of the lesser degree of leftward shift of axis is also likely to be higher. Even the lesser degree of leftward shift of axis is also quite significant in children as compared to adults, as the normal QRS axis is more on the right in children. For example, in the neonate the normal frontal plane QRS axis is between +60° and +190°, and the axis then shifts to the left and by ages 1–5 years, it is generally between +10° and +110°. Between 5 and 8 years of age, the normal QRS axis may extend to +140°, and between ages 8 and 16 years, the range of normal QRS extends to +120° [3]. The indigenous populations like the American‐Indian are the ones who are affected the most since the middle of the twentieth century by the global diabetes epidemic [47, 48], and there is high prevalence of diabetes in the American‐Indian indigenous population affecting even children [47, 51]. The high prevalence of leftward shift of axis in the children in such population is most likely related to the glucose intolerance which needs to be studied.

## **14. Brain white matter hyperintensities and left anterior fascicular block similarly related to glucose intolerance**

White matter of the brain consists mostly of glial cells and myelinated axons for the transmission of neuronal electrical activity. With the wide availability of magnetic resonance imaging, there is often incidental discovery of white matter lesions appearing as hyperintensities on T2 weighted image [52, 53]. Pathological findings in the regions of white matter hyperintensities include myelin pallor, tissue rarefaction associated with loss of myelin and axons, and mild gliosis [52]. The factors associated with the brain white matter hyperintensities include ageing, hypertension, and diabetes [52, 53], as in the case of left anterior fascicular block. It may be relevant to note here that neuropathy, including the autonomic one, is a well‐known common complication of diabetes and it may also be linked with, for example, as a later clinical manifestation of, white matter hyperintensities. A new study shows that even the impaired fasting glycemia, with the fasting plasma glucose below the diabetic range, is associated with a higher burden of brain white matter hyperintensities [54]. Among the people with isolated left axis deviation almost half of the persons less than 40 years of age have blood glucose in the upper quintile values [28]. Left anterior fascicle is similarly involved in the *transmission of cardiac electrical activity* and its vulnerability to the interruption (**Table 3**) could make it likely to be susceptible to oxidative injury due to the accumulation of various metabolic products of hyperglycemia.

Hyperglycemia is associated with fibrodegeneration of the left anterior fascicle, brain white matter, and other tissues. Chronic hyperglycemia affects various growth factors including fibroblast, collagen, fibronectin, contractile proteins, and extracellular matrix proteins in the body through different mechanisms. The various possible mechanisms of hyperglycemia leading to complications include nonenzymatic glycosylation, polyol pathway, abnormal microvascular blood flow, thickening and leakage of basement membrane of blood vessels, formation of reactive oxygen species, formation of vascular endothelial growth factors, and overproduction of superoxide by the mitochondrial electron chain [11, 12, 55]. Ageing associated with the fibrodegeneration of various tissues also involves insulin signaling pathways, reactive oxygen species, and oxidative damage at number of sites [55]. The final pathways of mechanism of complications and fibrodegeneration of various tissues due to hyperglycemia and ageing appear similar. Ageing is a known risk factor of glucose intolerance. And looking at the similar final pathways of mechanisms of complications and fibrodegeneration of the tissues in hyperglycemia and ageing, there could also be reciprocal relation between the two conditions.

## **15. Future perspectives—from research and public health point of views**

From the point of view of leftward shift of axis, correlation of various degrees of frontal plane QRS axis (not just the presence or absence of left axis deviation) with fasting and 2‐hour glucose and/or glycated hemoglobin levels


will help to provide further evidences and thus to correlate various factors with leftward shift of axis and glucose intolerance. The value of research lies in its utility. In the situation of pandemic of glucose intolerance also affecting the leftward shift of axis in younger population, to utilize the already available evidences especially of the risk of obesity or glucose intolerance in the offspring of mother with obesity or with undernutrition [56], the control programs to protect the susceptible populations need to be implemented [47] (**Table 4**). In the background of inherent insulin resistance during pregnancy and increasing age of mothers, maintenance of the optimum prepregnancy weight in the population appears to the key in hand in the control program of diabetes epidemic [47]. This will also help to benefit the children of the indigenous population where the research on left axis deviation and diabetes has been conducted.


\*The vulnerable populations to be protected by the control program of diabetes include the offspring of malnourished or overweight mothers.

\*\*National and international health and diabetes agencies should clearly spell out the control programs, with appropriate budget allocation, for control of diabetes epidemic to protect the progeny.

**Table 4.** Examples of prevention and control programs in the communicable diseases as a model for similar strategies for individuals and susceptible populations in diabetes epidemic.

## **16. Conclusion**

the upper quintile values [28]. Left anterior fascicle is similarly involved in the *transmission of cardiac electrical activity* and its vulnerability to the interruption (**Table 3**) could make it likely to be susceptible to oxidative injury due to the accumulation of various metabolic products of

Hyperglycemia is associated with fibrodegeneration of the left anterior fascicle, brain white matter, and other tissues. Chronic hyperglycemia affects various growth factors including fibroblast, collagen, fibronectin, contractile proteins, and extracellular matrix proteins in the body through different mechanisms. The various possible mechanisms of hyperglycemia leading to complications include nonenzymatic glycosylation, polyol pathway, abnormal microvascular blood flow, thickening and leakage of basement membrane of blood vessels, formation of reactive oxygen species, formation of vascular endothelial growth factors, and overproduction of superoxide by the mitochondrial electron chain [11, 12, 55]. Ageing associated with the fibrodegeneration of various tissues also involves insulin signaling pathways, reactive oxygen species, and oxidative damage at number of sites [55]. The final pathways of mechanism of complications and fibrodegeneration of various tissues due to hyperglycemia and ageing appear similar. Ageing is a known risk factor of glucose intolerance. And looking at the similar final pathways of mechanisms of complications and fibrodegeneration of the tissues in hyperglycemia and ageing, there could also be reciprocal relation between the

**15. Future perspectives—from research and public health point of views**

From the point of view of leftward shift of axis, correlation of various degrees of frontal plane QRS axis (not just the presence or absence of left axis deviation) with fasting and 2‐hour glu-

• in the people with higher brain white matter hyperintensities [52–54] (along with magnetic resonance imaging of heart and/or nerve conduction studies and/or tests of autonomic

will help to provide further evidences and thus to correlate various factors with leftward shift of axis and glucose intolerance. The value of research lies in its utility. In the situation of pandemic of glucose intolerance also affecting the leftward shift of axis in younger population, to utilize the already available evidences especially of the risk of obesity or glucose intolerance in the offspring of mother with obesity or with undernutrition [56], the control programs to protect the susceptible populations need to be implemented [47] (**Table 4**). In the background of inherent insulin resistance during pregnancy and increasing age of mothers, maintenance of the optimum prepregnancy weight in the population appears to the key in hand in the control program of diabetes epidemic [47]. This will also help to benefit the children of the indigenous population where the research on left axis deviation and diabetes has been conducted.

• in the children of the indigenous population where left axis is observed [50], and

hyperglycemia.

30 Interpreting Cardiac Electrograms - From Skin to Endocardium

two conditions.

neuropathy)

cose and/or glycated hemoglobin levels

• in the general population [45],

The frontal plane QRS axis and especially the left axis deviation have always been the areas of interest in electrocardiogram. There are different physiological and pathological conditions which affect the axis and the axis shift itself also affects the QRS and ST morphologies. The approximate degree of axis can be easily determined by observing the electrocardiogram. Most causes of left axis deviation are well‐known clinical entities. Isolated left axis deviation and leftward shift of the axis have increasingly been reported to be possibly associated with glucose intolerance. There are reasons why such association was previously not reported. The left anterior fascicle is as such vulnerable to interruption. The possible relation of glucose intolerance with brain white matter hyperintensities and even ageing also indicate the need to conduct research in these areas. However, the already available evidences should also be simultaneously utilized to protect the susceptible population. The bottom line of the frontal plane QRS axis is to record the actual degrees of the axes (not just the presence or absence of normal axis or left, right, or extreme axis deviations) and correlate the changes in the degrees of axis with the levels of the various possible factors in the individual patient or the study populations.

### **Author details**

Madhur Dev Bhattarai

Address all correspondence to: mdb@ntc.net.np

Nepal Diabetes Association, Kathmandu, Nepal

## **References**


[15] Raynaud, Valeix, Drouet, Escourrou, Durand. Electrocardiographic observations in high altitude residents of Nepal and Bolivia. International Journal of Biometerology. 1981;**25**(3):205‐217

**Author details**

**References**

2013

Madhur Dev Bhattarai

Address all correspondence to: mdb@ntc.net.np Nepal Diabetes Association, Kathmandu, Nepal

32 Interpreting Cardiac Electrograms - From Skin to Endocardium

ed. New Delhi: Wiley; 2013

[1] Dubin D. Rapid Interpretation of EKGs. 6th ed. Tampa: Cover Publishing Co; 2000

conduction disturbances. Circulation. 2009;**119**:e235‐e240

[4] Prutkin JM. Basic principles of ECG analysis. UpToDate. 2017, Feb

Wolters Kluwer Health/Lippincott Williams and Wilkins; 2014

Medical Publications/Maruzen Asia; 1982

Heart Association; 1990

pp. 944‐947, 1264‐1273

Simplified Approach. 8th ed. Philadelphia: Elsevier/Saunders; 2013

[2] Narasimhan C, Franchis J. Leo Shamroth—An Introduction to Electrocardiography. 8th

[3] Surawicz B, Childers R, Deal BM, Gettes LS. AHA/ACCF/HRS recommendations for the standardization and interpretation of the electrocardiogram Part III: Intraventricular

[5] Goldberger Z, Shvilkin A, Goldberger AL. Goldberger's Clinical Electrocardiography—A

[6] Wegner GS, Strauss DG. Marriott's Practical Electrocadiography. 12th ed. New Delhi:

[7] Hampton JR. The ECG Made Easy. 8th ed. Edinburg: Elsevier/Churchill Livingstone;

[8] Goldman MJ. Principles of Clinical Electrocardiography. 11th ed. California: Lange

[9] Akhtar M (Prepared on behalf of the Council on Clinical Cardiology of the American Heart Association). Examination of the Heart—The Electrocardiogram. Dallas: American

[10] Blackburn H, Keys A, Simonson E, Rautharju P, Punsar S. The electrocardiographic in

[11] Kumar, Clark. Kumar and Clark's Clinical Medicine. 9th ed. Edinburgh: Elsevier; 2017.

[12] Waugh A, Grant A. Ross and Wilson's Anatomy and Physiology in Health and Illness.

[14] Pryour, Weaver, Blount. Electrocardiographic observations of 493 residents living at high altitude (10,150 feet). American Journal of Cardiology. 1965;**16**(4):494‐499

population studies—A classification system. Circulation. 1960;**21**:1160‐1175

11th ed. Edinburgh: Churchill Livingstone/Elesevier; 2010. pp. 80‐83

[13] Sauer WH. Left anterior fascicular block. UpToDate. 2017


[44] Paudyal A (under the guidance of Bhattarai MD). Correlation of normal QRS duration left axis deviation in ECG with clinical and investigation parameters in patients without cardiac symptoms [thesis for MD in Internal Medicine]. Kathmandu: National Academy of Medical Sciences; 2008

[29] Kitchin AH, Lowther CP, Milne JS. Prevalence of clinical and electrocardiographic evidence of ischaemic heart disease in the older population. British Heart Journal.

[30] de Bacquer D, de Baker G, Kornitzer M. Prevalences of ECG findings in large population

[31] Lakkireddy DR, Clark RA, Mohiuddin SM. Electrocardiographic findings in patients>100 years of age without clinical evidence of cardiac disease. The American Journal of

[32] Ostrander LD, Brandt RL, Kjelsberg MO, Epstein FH. Electrocardiographic findings among the adult population of a natural community, Tecumseh, Michigan. Circulation.

[33] Hingorani P, Natekar M, Deshmukh S, Karnad DR, Kothari S, Narula D, Lokhandwala Y. Morphological abnormalities in baseline ECGs in healthy normal volunteers participating in phase I studies. The Indian Journal of Medical Research. 2012;**135**:322‐330 [34] Hiss RG, Lamb LE. Electrocardiographic findings in 122,043 individuals. Circulation.

[35] Mason JW, Ramseth DJ, Chanter DO, Moon TE, Goodman DB, Mendzelevski B. Electrocardiographic reference ranges derived from 79,743 ambulatory subjects. Journal

[36] Bahl OP, Walsh TJ, Massie E. Left axis deviation: An electrocardiographic study with

[37] Grayzel J, Neyshaboori M, Paramw NJ. Left‐axis deviation: Etiologic factors in one‐hun-

[38] Corne RA, Beasmish RE, Rollwagen RL. Significance of left anterior hemiblock. British

[39] Grant RP. Left axis deviation: An electrocardiographic‐pathological correlation study.

[40] Das G. Left axis deviation—A spectrum of intraventricular conduction block. Circulation.

[41] Bradlow BA. The importance of abnormal left axis deviation in life assurance. South

[42] van Campenhout MJH, Yaksh A, Kik C, de Jaegere PP, Yen S, Allessie MA, de Groot NMS. Bachmann's Bundle—A key player in the development of atrial fibrillation.

[43] Uusitupa M, Mustonen J, Siitonen O, Pyorala K. Quantitative electrocardiographic and vectorcardiographic study on newly‐diagnosed non‐insulin‐dependent diabetic and

post‐mortem correlation. British Heart Journal. 1969;**31**(4):451‐456

Circulation Arrhythmia and Electrophysiology. 2013;**6**:1041‐1046

non‐diabetic control subjects. Cardiology. 1988;**75**(1):1‐9

dred patients. American Heart Journal. 1975;**89**(4):419‐427

based samples of men and women. Heart. 2000;**84**(6):625‐633

1973;**35**(9):946‐953

1965;**31**:888‐898

1962;**25**:947‐961

Cardiology. 2003;**92**(10):1249‐1251

34 Interpreting Cardiac Electrograms - From Skin to Endocardium

of Electrocardiology. 2007;**40**(3):228‐234

Heart Journal. 1978;**40**(5):552‐557

Circulation. 1956;**14**(2):233‐249

African Medical Journal. 1973:**47**(20):877‐881

1976;**53**(6):917‐919


**Provisional chapter**

## **Signal-Averaged ECG: Basics to Current Issues**

**Signal-Averaged ECG: Basics to Current Issues**

DOI: 10.5772/intechopen.69279

Ioana Mozos and Dana Stoian Ioana Mozos and Dana Stoian Additional information is available at the end of the chapter

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/intechopen.69279

#### **Abstract**

Signal-averaged ECG (SAECG) is a high-resolution, noninvasive electrocardiographic method enabling detection of late ventricular potentials (LVP), which are low-amplitude and high-frequency signals, predicting reentry ventricular arrhythmias, and sudden cardiac death (SCD). Three criteria are used to detect late ventricular potentials as follows: signal-average ECG QRS duration (SAECG-QRS), the duration of the terminal part of the QRS complex with an amplitude below 40 μV (LAS40) and the root mean square (RSM) signal amplitude of the last 40 ms of the signal < 20 μV (RMS40). Late ventricular potentials can be detected not only at the end of a QRS complex but also as intra-QRS (IQRS) potentials. Signal-averaged ECG was modified to enable the analysis of the P-wave and to detect atrial late potentials (ALPs), low-amplitude potentials at the terminal part of the filtered P-wave, and predictors of atrial fibrillation (AF). Late atrial and ventricular potentials originate from areas of delayed, fragmented, and heterogenous conduction within atrial or ventricular myocardium. This chapter reviews the most important mechanisms explaining the occurrence of late ventricular, intra-QRS, and atrial potentials; their predictive value for arrhythmia, focusing on recent clinical data, long-term follow-up, and outcome; and analysis of SAECG variables in cardiac and noncardiac diseases.

**Keywords:** late ventricular potentials, atrial late potentials, ventricular arrhythmia risk, atrial fibrillation

## **1. Introduction**

Cardiovascular disorders are leading mortality causes worldwide. Prophylactic methods and early detection deserve special attention. The standard 12-lead electrocardiogram (ECG) is a simple, reliable, and cost-effective method, used in clinical practice and trials for sudden cardiac death (SCD) risk stratification, considering QT and QRS duration, fragmented QRS complexes, and Tpeak-tend interval. Signal-averaged ECG (SAECG) can detect very small, subtle signals (microvolt level), which are not visible when using standard 12-lead ECG, by

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2017 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

averaging and filtering multiple ECG complexes [1–3]. The high-resolution or signal-averaged ECG has been recommended by the European Society of Cardiology, the American Heart Association, and the American College of Cardiology as a useful tool to improve the diagnosis and risk stratification of patients with ventricular arrhythmias or those at risk of developing life-threatening ventricular arrhythmias [4].

The substrate for SCD varies from advanced cardiomyopathic injuries, myocardial infarction scars to no obvious sign of structural damage [4]. The most common cause of SCD is coronary heart disease, but several cardiomyopathies, heart failure, and genetic influences, as well as myocarditis, pericardial diseases, pulmonary arterial hypertension, rheumatic disease, endstage renal failure, endocrine disorders, obesity, anorexia, hypertension, lipid abnormalities, diabetes mellitus, several drugs, and physical and toxic agents can also be involved [4]. Several inherited abnormalities, including long and short QT interval, Brugada syndrome, and catecholaminergic ventricular tachycardia (VT), can precipitate SCD without any structural changes in the heart, triggered by external events [4].

Atrial fibrillation (AF) is the most frequent arrhythmia in the general population, with poorly understood underlying mechanisms of structural and electrical atrial remodeling [5]. It is associated with an increased risk of stroke, heart failure, and mortality [5].

The aim of this chapter is to review the most important mechanisms explaining the occurrence of late ventricular, intra-QRS (IQRS), and atrial potentials; their predictive value for arrhythmia, focusing on recent studies, long-term follow-up, and outcome; and analysis of SAECG variables in cardiac and noncardiac diseases.

## **2. Late ventricular potentials**

LVPs are low-amplitude, high-frequency signals, occurring in the terminal part of the QRS complex, as markers of electrophysiological cardiac substrates for reentry ventricular arrhythmia, favored by structural heterogeneity due to myocardial necrosis, fibrosis, or dystrophy [6]. LVPs appear if conduction is slow enough to enable reentry and a unidirectional block is present [6]. They assess ventricular depolarization, and the signal is more stable and reproducible than the repolarization process [6]. Arrhythmia triggers are autonomic imbalances (increased sympathetic activity), acute ischemia, or electrolyte disorders. Temporal and frequency domain analysis can be performed to detect arrhythmia risk.

Three criteria are used to detect LVPs as follows: SAECG-QRS duration , the duration of the terminal part of the QRS complex with an amplitude below 40 μV (LAS40), and the root mean square signal amplitude of the last 40 ms of the signal < 20 μV (RMS40) [7].

SAECG-QRS was considered prolonged if it exceeds 120 [7] or 114 ms according to other authors [8, 9]. LAS40 is pathological if exceeding 38 ms and RMS40 if less than 20 μV [7]. Late ventricular potentials are defined by the presence of one or two of the mentioned positive criteria (**Figures 1**, **2**) [7, 10].

averaging and filtering multiple ECG complexes [1–3]. The high-resolution or signal-averaged ECG has been recommended by the European Society of Cardiology, the American Heart Association, and the American College of Cardiology as a useful tool to improve the diagnosis and risk stratification of patients with ventricular arrhythmias or those at risk of developing

The substrate for SCD varies from advanced cardiomyopathic injuries, myocardial infarction scars to no obvious sign of structural damage [4]. The most common cause of SCD is coronary heart disease, but several cardiomyopathies, heart failure, and genetic influences, as well as myocarditis, pericardial diseases, pulmonary arterial hypertension, rheumatic disease, endstage renal failure, endocrine disorders, obesity, anorexia, hypertension, lipid abnormalities, diabetes mellitus, several drugs, and physical and toxic agents can also be involved [4]. Several inherited abnormalities, including long and short QT interval, Brugada syndrome, and catecholaminergic ventricular tachycardia (VT), can precipitate SCD without any struc-

Atrial fibrillation (AF) is the most frequent arrhythmia in the general population, with poorly understood underlying mechanisms of structural and electrical atrial remodeling [5]. It is

The aim of this chapter is to review the most important mechanisms explaining the occurrence of late ventricular, intra-QRS (IQRS), and atrial potentials; their predictive value for arrhythmia, focusing on recent studies, long-term follow-up, and outcome; and analysis of

LVPs are low-amplitude, high-frequency signals, occurring in the terminal part of the QRS complex, as markers of electrophysiological cardiac substrates for reentry ventricular arrhythmia, favored by structural heterogeneity due to myocardial necrosis, fibrosis, or dystrophy [6]. LVPs appear if conduction is slow enough to enable reentry and a unidirectional block is present [6]. They assess ventricular depolarization, and the signal is more stable and reproducible than the repolarization process [6]. Arrhythmia triggers are autonomic imbalances (increased sympathetic activity), acute ischemia, or electrolyte disorders. Temporal and

Three criteria are used to detect LVPs as follows: SAECG-QRS duration , the duration of the terminal part of the QRS complex with an amplitude below 40 μV (LAS40), and the root mean

SAECG-QRS was considered prolonged if it exceeds 120 [7] or 114 ms according to other authors [8, 9]. LAS40 is pathological if exceeding 38 ms and RMS40 if less than 20 μV [7]. Late

life-threatening ventricular arrhythmias [4].

38 Interpreting Cardiac Electrograms - From Skin to Endocardium

tural changes in the heart, triggered by external events [4].

SAECG variables in cardiac and noncardiac diseases.

**2. Late ventricular potentials**

associated with an increased risk of stroke, heart failure, and mortality [5].

frequency domain analysis can be performed to detect arrhythmia risk.

square signal amplitude of the last 40 ms of the signal < 20 μV (RMS40) [7].

Considering their low amplitude, LVPs can only be detected if amplified, filtered, and averaged using high-resolution SAECG or body surface mapping. Electronic filters can further reduce signal noise by eliminating high-frequency signals such as skeletal muscle potentials [3]. The filters used in SAECG provide different numerical and diagnostic results, with a higher sensitivity for 40–250-Hz filters compared to 40-Hz filters [1]. The authors of the present chapter have experience only with 50–250 Hz filters. ECG signals are collected for 5–20 min, followed by averaging the QRS complexes through the temporal technique to reduce the signal-to-noise ratio [3].

The most important limitations in SAECG are related to electrical interference causing false results and the low positive predictive value for arrhythmic events [6, 11, 12]. However, they have a high negative predictive value for arrhythmic events [12]. Their presence predicts inducibility of ventricular tachycardia at invasive electrophysiology studies, and if they are combined with low ejection fraction, they enable detection of patients at high risk of sudden cardiac death [3].

Besides time-domain (TD) analysis of SAECG, frequency domain analysis may also provide valuable data. Abrupt changes in the frequency contents between adjacent overlapping segments of the QRS complex are the markers of the arrhythmogenic substrate in spectral turbulent analysis (STA) [13].

**Figure 1.** Late ventricular potentials with two positive criteria (LAS40 and RMS40).

**Figure 2.** Normal SAECG and no late ventricular potentials.

The sensitivity of SAECG is higher compared to standard 12-lead ECG for identifying patients with acute coronary syndrome [14]. LVPs were initially used in patients with myocardial infarction. They appear in the heterogenous tissue at the border of a myocardial infarction scar [6], very frequent in nonseptal myocardial segments, and were abolished in most patients with myocardial infarction after ventricular tachycardia ablation, associated with scar homogenization and a low recurrence rate [15]. The utility of SAECG was questioned in the postpercutaneous coronary intervention era [3]. LVPs have been recorded in several other cardiac disorders, especially cardiomyopathies, myocarditis, infiltrative heart disease, arrhythmogenic right ventricular dysplasia, congenital heart defects, heart failure, left ventricular hypertrophy, Brugada syndrome, early repolarization, bundle branch block, and atrial fibrillation [6, 16–18]. Despite improved postinfarction survival due to lifestyle changes, thrombolytic, antiplatelet therapy, beta-blockers, and revascularization, LVPs can still be used in selecting patients for interventional studies [6]. Dinov et al. [19] found a positive correlation between endocardial scar area and filtered QRS in patients with ischemic VT, normalization of SAECG after catheter ablation (CA), and abnormal SAECG after CA as a predictor for VT recurrence (**Table 1**). Conduction delay contributed to ventricular dyssynchrony, regardless of LVPs in patients with heart failure, and LVPs did not play an important role in ventricular dyssynchrony [16]. Several SAECG studies have been performed in patients who underwent heart transplant [20–22]. SAECG distinguished between heart transplant patients with or without rejection, especially LAS40 and RMS40 [22]. The association between LVPs and rejection of heart transplant is explained by occurrence of areas of myocardial fibrosis, due to cell changes caused by alloreactive T lymphocytes against graft antigens and ischemia-reperfusion injuries as soon as the blood flow is reestablished [22].


The sensitivity of SAECG is higher compared to standard 12-lead ECG for identifying patients with acute coronary syndrome [14]. LVPs were initially used in patients with myocardial infarction. They appear in the heterogenous tissue at the border of a myocardial infarction scar [6], very frequent in nonseptal myocardial segments, and were abolished in most patients with myocardial infarction after ventricular tachycardia ablation, associated with scar homogenization and a low recurrence rate [15]. The utility of SAECG was questioned in the postpercutaneous coronary intervention era [3]. LVPs have been recorded in several other cardiac disorders, especially cardiomyopathies, myocarditis, infiltrative heart disease, arrhythmogenic right ventricular dysplasia, congenital heart defects, heart failure, left ventricular hypertrophy, Brugada syndrome, early repolarization, bundle branch block, and atrial fibrillation [6, 16–18]. Despite improved postinfarction survival due to lifestyle changes, thrombolytic, antiplatelet therapy, beta-blockers, and revascularization, LVPs can still be used in selecting patients for interventional studies [6]. Dinov et al. [19] found a positive correlation between endocardial scar area and filtered QRS in patients with ischemic VT, normalization of SAECG after catheter ablation (CA), and abnormal SAECG after CA as a predictor for VT recurrence (**Table 1**). Conduction delay contributed to ventricular dyssynchrony, regardless of LVPs in patients with heart failure, and LVPs did not play an important role in ventricular dyssynchrony [16]. Several SAECG studies have been performed in patients who underwent heart transplant [20–22]. SAECG distinguished between heart transplant patients with or without rejection, especially LAS40 and RMS40 [22]. The association between LVPs and rejection of heart transplant is explained by occurrence of areas of myocardial fibrosis, due to cell changes caused by alloreactive T lymphocytes against graft antigens and ischemia-reperfusion injuries

as soon as the blood flow is reestablished [22].

**Figure 2.** Normal SAECG and no late ventricular potentials.

40 Interpreting Cardiac Electrograms - From Skin to Endocardium


**Table 1.** LVP studies.

Extracardiac disorders were also associated with LVPs, especially hypertension, metabolic syndrome, obesity, eating disorders, diabetes mellitus, renal failure, chronic obstructive pulmonary disease (COPD), acromegaly, thalassemia, connective tissue diseases, epilepsy, and schizophrenia [6, 23–27]. Antiarrhythmic therapy, thrombolytic drugs, statins, steroids, and coronary interventions may influence LVPs [6].

Sudden cardiac death is higher in psychiatric patients, especially those with depression and schizophrenia than in the general population [28]. Several factors influence the relationship with cardiovascular disorders in patients with depression: social factors (poverty, social inequality, reduced access to healthcare), biological factors (endothelial dysfunction, impaired heart rate variability and platelet function, inflammation, hyperactivity of hypothalamic-pituitary-adrenal axis), higher prevalence of cardiovascular risk factors, and therapy (side effects of tricyclic, lower adherence) [29]. Both schizophrenia and depression impair the autonomic tone, ion channels, alter connexin 43 expression, and may cause drug-induced cardiac fibrosis [28].

Several factors enable ventricular arrhythmias in patients with epilepsy, such as sympathovagal imbalance, impaired cardiac repolarization, mutations of ionic channels affecting both the brain and the heart, dysfunctional cortical networks, ictal hypoxemia and hypercapnia, stress hormones, therapy, cardiorespiratory interactions, and associated cardiovascular diseases [24, 30]. Epilepsy patients more frequently displayed abnormal SAECGs with LVPs compared to healthy controls, correlated with disease duration, uncontrolled seizures, and polytherapy [23]. Svalheim et al. [31] reported no electrocardiographic changes (in standard ECG and SAECG) after antiepileptic drugs (carbamazepine and lamotrigene) in 26 epileptic patients.

COPD was associated with cardiovascular morbidity and mortality, considering negative cardiac effects of hyperinflation, exercise limitation, smoking, and hypoxemia [6]. Carjea found a higher prevalence of LVPs in patients with COPD, especially in moderate and severe cases [32]. Yildiz et al. [33] reported a significantly increased total QRS duration in patients with COPD compared to control subjects and LVPs but no significant association with premature ventricular contractions. Despite higher prevalence of LVPs, premature ventricular contractions, and complex ventricular arrhythmias in patients with COPD compared to healthy controls, SAECG had little value in stratification of ventricular arrhythmia risk in a study including 41 patients with COPD and 63 patients without any history of pulmonary disease [27].

Persistent, life-threatening ventricular arrhythmias may occur in several endocrine disorders, such as pheochromocytoma, acromegaly, primary aldosteronism, Addison disease, hypo- and hyperparathyroidism, and hypothyroidism [4]. Ventricular arrhythmias may occur due to excess or insufficient hormone activity on myocardial receptors, myocardial changes, electrolyte imbalances, or acceleration of progression of structural cardiac disorders [4]. Sudden death and increased prevalence of ventricular arrhythmias and LVPs have been described in acromegaly [34]. Ventricular arrhythmia risk in acromegaly is related to the specific cardiomyopathy associated with left ventricular hypertrophy, myocardial fibrosis, comorbidities, especially hypertension and sleep apnea, and, possibly, to the direct effects of the growth hormone and insulin growth factor 1 on myocardial cells and cardiac ion channels [34, 35]. The prevalence of LVPs was significantly higher in patients with acromegaly compared to healthy controls, related to a longer duration of the disease, premature ventricular contractions, and left ventricular hypertrophy [34]. Herrmann et al. [36] also reported LVPs in patients with active and well-controlled acromegaly, as a sensitive and early sign of myocardial injury, not related to muscle mass and body mass index, age, gender, and duration of the disease.

Thyroid hormone exerts several effects on the cardiovascular system [37]. Ventricular arrhythmia and sudden cardiac death may occur especially in hypothyroidism, probably related to prolonged QT interval [4]. LVPs have been described in hypo- and hyperthyroidism, according to a study including 278 patients with thyroid disorders even in subclinical dysfunctions [38]. A case of severe primary hypothyroidism was presented with an abnormal SAECG with LVPs, which disappeared with thyroxine therapy [37].

Future SAECG studies should also include patients with Cushing's syndrome, considering impaired cardiac function and structure due to the direct toxic effect of cortisol, increased blood pressure, central obesity, metabolic syndrome, hyperglycemia, and chronic hypokalemia [39]. Subclinical structural and functional cardiac alterations are very common but underdiagnosed [39].

## **3. Intra-QRS potentials**

Extracardiac disorders were also associated with LVPs, especially hypertension, metabolic syndrome, obesity, eating disorders, diabetes mellitus, renal failure, chronic obstructive pulmonary disease (COPD), acromegaly, thalassemia, connective tissue diseases, epilepsy, and schizophrenia [6, 23–27]. Antiarrhythmic therapy, thrombolytic drugs, statins, steroids, and

**Study population Results Follow up References**

Maffei et al. [34]

Herrmann et al. [36]

A higher prevalence of LVPs in acromegaly which significantly correlated with Lown scale of premature ventricular

LVPs are frequently seen in active acromegaly as an early and sensitive parameter of myocardial

contractions

injury

Sudden cardiac death is higher in psychiatric patients, especially those with depression and schizophrenia than in the general population [28]. Several factors influence the relationship with cardiovascular disorders in patients with depression: social factors (poverty, social inequality, reduced access to healthcare), biological factors (endothelial dysfunction, impaired heart rate variability and platelet function, inflammation, hyperactivity of hypothalamic-pituitary-adrenal axis), higher prevalence of cardiovascular risk factors, and therapy (side effects of tricyclic, lower adherence) [29]. Both schizophrenia and depression impair the autonomic tone, ion channels, alter con-

Several factors enable ventricular arrhythmias in patients with epilepsy, such as sympathovagal imbalance, impaired cardiac repolarization, mutations of ionic channels affecting both the brain and the heart, dysfunctional cortical networks, ictal hypoxemia and hypercapnia, stress hormones, therapy, cardiorespiratory interactions, and associated cardiovascular diseases [24, 30]. Epilepsy patients more frequently displayed abnormal SAECGs with LVPs compared to healthy controls, correlated with disease duration, uncontrolled seizures, and polytherapy [23]. Svalheim et al. [31] reported no electrocardiographic changes (in standard ECG and SAECG) after antiepileptic drugs (carbamazepine and lamotrigene) in 26 epileptic patients.

COPD was associated with cardiovascular morbidity and mortality, considering negative cardiac effects of hyperinflation, exercise limitation, smoking, and hypoxemia [6]. Carjea found a higher prevalence of LVPs in patients with COPD, especially in moderate and severe cases [32]. Yildiz et al. [33] reported a significantly increased total QRS duration in patients with COPD compared to control subjects and LVPs but no significant association with premature ventricular contractions. Despite higher prevalence of LVPs, premature ventricular contractions, and complex ventricular arrhythmias in patients with COPD compared to healthy controls, SAECG had little value in stratification of ventricular arrhythmia risk in a study including 41

patients with COPD and 63 patients without any history of pulmonary disease [27].

coronary interventions may influence LVPs [6].

70 acromegalic patients and 70 control subjects, age- and

48 patients with acromegaly: 16 active disease, 32 cured or 'well controlled', under treatment with sandostatin analogs, and 38 healthy volunteers

42 Interpreting Cardiac Electrograms - From Skin to Endocardium

sex-matched

**Table 1.** LVP studies.

nexin 43 expression, and may cause drug-induced cardiac fibrosis [28].

IQRSPs are low-amplitude notches (the order of microvolts), usually invisible in the standard ECG, which may occur anywhere in the signal-averaged QRS [2] and may not prolong the normal QRS duration [40]. They were described as the signals with sudden slope changes [40]. Extracting IQRSPs is challenging, considering that they are very weak signals, with abrupt changes in slope, approximation errors, and the differences among patients with ventricular arrhythmias [41]. The root mean square values were highly correlated with the parameters of the abnormal intra-QRS potentials in healthy controls but not in patients with ventricular tachycardia [40].

A combination of IQRSPs and LVPs can improve predictive accuracy for patients of high risk of ventricular arrhythmias.

## **4. P-wave potentials**

P-wave signal-averaged electrocardiography, atrial late potentials (ALP), and abnormal intra-P-wave potentials could detect patients at risk of supraventricular arrhythmias, especially atrial fibrillation [42, 43]. ALP originates from areas of delayed and heterogenous conduction within the atrial myocardium, responsible for the occurrence of AF [44].

Prolonged filtered P-wave duration (FPD) in P-wave signal-averaged electrocardiography has been used as a noninvasive, powerful predictor of AF, the first episode and recurrences, in lone, occult or silent atrial fibrillation, in stroke, heart failure, hypertension, hypertrophic cardiomyopathy, hypothyroidism and in patients undergoing coronary artery bypass surgery [44–46]. A prolonged SAECG P-wave duration was also mentioned in septal atrial defect, especially in patients who experienced AF, not corrected after atrial septal defect closure, and it was demonstrated that atrial conduction disturbances occur early, requiring an early intervention to prevent the development of late AF (**Table 2**) [47].

There is no consensus about the cut-off point for FPD, which was 121 ms in hypertensive patients Auriti et al. [48], 124 ms in patients in sinus rhythm, 136 ms in hypertensive patients with a history of atrial fibrillation, 132 ms in patients with COPD, and 155 ms in several other studies [43, 45, 46, 49], differences related to different averaging and filtering methods [45].



**Table 2.** Atrial late potentials (ALP) and fragmented electrical activity on the P-wave.

Besides FPD, Buzea et al. [43] also used the RMS voltages in the last 40, 30, and 20 ms of the filtered P-wave (RMS 40, RMS 30, and RMS 20), the root mean square voltage of the filtered P-wave potentials (RMS-p), and the integral of the potentials during the filtered P-wave (Integral-p), and defined ALP as FPD > 132 ms and RMS 20 < 2.3 μV.

Fragmentations are expected to occur throughout atrial depolarization and not only in its terminal part, and inter- and intra-atrial conduction may be impaired [42]. High-frequency fragmented electrical activity on the P-wave in patients with recurrent AF is the expression of atrial electrical heterogeneity, responsible for reentry circuits in the atria [42]. Barbosa et al. [42] used spectral turbulence analysis of the P-SAECG to detect abnormal intra-P-wave potentials, demonstrating that fragmented electrical activity is an independent predictor of early AF recurrence.

## **5. Limitations**

**4. P-wave potentials**

44 Interpreting Cardiac Electrograms - From Skin to Endocardium

45 patients with exacerbation of COPD and 58 age-matched patients with no history of pulmonary disease in a control group

37 hypertensive patients with a first

68 stroke patients in sinus rhythm,

37 age- and sex-matched hypertensive controls without AF

without history of AF

35 patients with atrial septal defect

4 generations kindred of 27 individuals, 8 with AF on the ECG

41 patients with two or more symptomatic episodes of idiopathic and persistent atrial fibrillation after successful electrical cardioversion

and 25 healthy controls

AF episode

P-wave signal-averaged electrocardiography, atrial late potentials (ALP), and abnormal intra-P-wave potentials could detect patients at risk of supraventricular arrhythmias, especially atrial fibrillation [42, 43]. ALP originates from areas of delayed and heterogenous conduction

Prolonged filtered P-wave duration (FPD) in P-wave signal-averaged electrocardiography has been used as a noninvasive, powerful predictor of AF, the first episode and recurrences, in lone, occult or silent atrial fibrillation, in stroke, heart failure, hypertension, hypertrophic cardiomyopathy, hypothyroidism and in patients undergoing coronary artery bypass surgery [44–46]. A prolonged SAECG P-wave duration was also mentioned in septal atrial defect, especially in patients who experienced AF, not corrected after atrial septal defect closure, and it was demonstrated that atrial conduction disturbances occur early, requiring an early intervention to

There is no consensus about the cut-off point for FPD, which was 121 ms in hypertensive patients Auriti et al. [48], 124 ms in patients in sinus rhythm, 136 ms in hypertensive patients with a history of atrial fibrillation, 132 ms in patients with COPD, and 155 ms in several other studies [43, 45, 46, 49], differences related to different averaging and filtering methods [45].

**Study population Results Follow up References**

Isolated atrial premature beats (APB) and supraventricular tachycardia (SVT) Buzea et al. [43]

Dakos et al. [50]

et al. [44]

Thilen et al. [47]

Darbar et al. [49]

Barbosa et al. [42]

11±4 months Yodogawa

8±6 months after atrial septal defect

closure

6 months, 12 months, atrial fibrillation recurrences

The patients with acute exacerbation of COPD have a higher incidence of supraventricular arrhythmias. P-wave SAECG analysis has little value in the arrhythmic risk evaluation of these

P-wave temporal and energy

ALP is a novel predictor of AF in stroke patients. P-SAECG should be

Prolonged P-wave duration does not change after atrial septal defect closure

Persons with AF and mutation carriers (on chromosome 5p15) can be identified by a prolonged P-SAECG duration

Fragmented electrical activity, use of amiodarone, and positive terminal portion of the Z-lead of the P-SAECG were independent predictors of recurrence of idiopathic and persistent atrial fibrillation

considered in stroke

characteristics can identify hypertensive patients at risk of AF recurrence

patients

within the atrial myocardium, responsible for the occurrence of AF [44].

prevent the development of late AF (**Table 2**) [47].

Most of the reviewed studies were observational, retrospective, with a low sample size and event rate, but careful statistical analysis may compensate the mentioned limitations. LVPs were detected using various equipment, commercially available or not, using different averaging and filtering methods. On the other hand, filtered QRS duration was not measured sequentially, considering therapy in all studies, and there was a lack of uniformity of the normality criteria for the diagnosis of LVPs.

False positive LVPs were reported in patients with junctional rhythm with retrograde P-waves, atrial flutter, and incomplete bundle branch block [51–53]. Combined TD and spectral turbulence analysis of the SAECG could improve its predictive value for fatal arrhythmias [54]. The positive predictive accuracy nearly doubled compared to TD or STA, without loss in sensitivity and specificity [54]. A high number of false positive LVPs was reported in myocardial infarction, as well, and in the early postinfarction period; in inferior myocardial infarction in time-domain analysis and anterior myocardial infarction according to STA [54, 55]. Delayed terminal conduction may increase the incidence of false positive results in SAECG, but the incidence of false positive LVPs was significantly lower if the combination of SAECG-QRS, LAS40, and RMS40 was used in patients with incomplete bundle branch block [53]. LVPs detected during sinus rhythm and lost after premature ventricular contractions may be responsible for false positive LVPs, and those revealed by ventricular extrastimuli and concealed during sinus rhythm may cause false negative LVPs [56]. Sensitivity might be low in patients with ventricular tachycardia due to early activation of potential sites of ventricular tachycardia in sinus rhythm, falling within the normal QRS duration [56]. The number of false positive results may be reduced by signal-averaging during premature ventricular stimulation [56].

Larger follow-up studies are needed to confirm the significance and usefulness of LVPs in different cardiac and noncardiac disorders.

#### **6. Conclusions**

This chapter brought back into focus SAECG, a noninvasive, low-cost, simple, and rapid method as a predictor of sudden cardiac death, using amplified ECG signals. Even though SAECG is not a routine screening test for sudden cardiac death risk and despite its low positive predictive value for arrhythmic events, LVPs and intra-QRS potentials provide valuable information not only in cardiac but also in extracardiac disorders, including psychiatric disorders, epilepsy, chronic obstructive pulmonary disease, and endocrine disorders. P-wave signal-averaged electrocardiography predicts atrial fibrillation episodes in patients with several disorders, such as hypertension, atrial septal defect, stroke, and chronic obstructive pulmonary disease.

#### **Author details**

Ioana Mozos<sup>1</sup> \* and Dana Stoian2

\*Address all correspondence to: ioanamozos@umft.ro

1 Department of Functional Sciences, Victor Babeș University of Medicine and Pharmacy, Timișoara, Romania

2 Department of Internal Medicine, Victor Babeș University of Medicine and Pharmacy, Timișoara, Romania

#### **References**


In: 16th IFAC Symposium on System Identification, the International Federation of Automatic Control; 11-13 July 2012; Brussels, Belgium, published by Elsevier Ltd.

[3] Abdelghani SA, Rosenthal TM, Morin DP. Surface electrocardiogram predictors of sudden cardiac arrest. Ochsner Journal. 2016;**16**:280-289

may be responsible for false positive LVPs, and those revealed by ventricular extrastimuli and concealed during sinus rhythm may cause false negative LVPs [56]. Sensitivity might be low in patients with ventricular tachycardia due to early activation of potential sites of ventricular tachycardia in sinus rhythm, falling within the normal QRS duration [56]. The number of false positive results may be reduced by signal-averaging during premature ven-

Larger follow-up studies are needed to confirm the significance and usefulness of LVPs in

This chapter brought back into focus SAECG, a noninvasive, low-cost, simple, and rapid method as a predictor of sudden cardiac death, using amplified ECG signals. Even though SAECG is not a routine screening test for sudden cardiac death risk and despite its low positive predictive value for arrhythmic events, LVPs and intra-QRS potentials provide valuable information not only in cardiac but also in extracardiac disorders, including psychiatric disorders, epilepsy, chronic obstructive pulmonary disease, and endocrine disorders. P-wave signal-averaged electrocardiography predicts atrial fibrillation episodes in patients with several disorders, such as hypertension, atrial septal defect, stroke, and chronic obstructive pul-

1 Department of Functional Sciences, Victor Babeș University of Medicine and Pharmacy,

2 Department of Internal Medicine, Victor Babeș University of Medicine and Pharmacy,

[1] Barbosa PRB, Barbosa EC, Bomfim AS, et al. Clinical assessment of the effect of digital filtering on the detection of ventricular late potentials. Brazilian Journal of Medical and Biological Research. 2002;**35**(11):1285-1292. DOI: 10.1590/S0100-879X2002001100005 [2] Ramos JA, Lopes dos Santos PJ. Parametric modeling in estimating abnormal intra-QRS potentials in signal-averaged electrocardiograms: A subspace identification approach.

tricular stimulation [56].

**6. Conclusions**

monary disease.

**Author details**

Timișoara, Romania

Timișoara, Romania

**References**

Ioana Mozos<sup>1</sup>

different cardiac and noncardiac disorders.

46 Interpreting Cardiac Electrograms - From Skin to Endocardium

\* and Dana Stoian2

\*Address all correspondence to: ioanamozos@umft.ro


[24] Mozos I. Ventricular arrhythmia risk in noncardiac diseases. In: Aronow WS, editor. Cardiac Arrhythmias—Mechanisms, Pathophysiology, and Treatment. InTech; 2014. DOI: 10.5772/57164. Available from: http://www.intechopen.com/books/cardiacarrhythmias-mechanisms-pathophysiology-and-treatment/ventricular-arrhythmiarisk-in-noncardiac-diseases

[13] Gottfridsson C, Karlsson T, Edvardsson N. The signal-averaged electrocardiogram before and after electrical cardioversion of persistent atrial fibrillation—implications of the sudden change in rhythm. Journal of Electrocardiology. 2011;**44**:242-250. DOI:

[14] Leisy PJ, Coeytaux RR, Wagner GS, et al. ECG-based signal analysis technologies for evaluating patients with acute coronary syndrome: A systematic review. Journal of

[15] Tsiachris D, Silberbauer J, Maccabelli G, et al. Electroanatomical voltage and morphology characteristics in postinfarction patients undergoing ventricular tachycardia ablation pragmatic approach favoring late potentials abolition. Circulation: Arrhythmia and

[16] Tahara T, Sogou T, Suezawa C, et al. Filtered QRS duration on signal-averaged electrocardiography correlates with ventricular dyssynchrony assessed by tissue Doppler imaging in patients with reduced ventricular ejection fraction. Journal of Electrocardiology.

[17] Ohkubo K, Watanabe I, Okumara Y, et al. Analysis of the spatial and transmural dispersion of repolarization and late potentials derived using signal-averaged vector-projected 187-channel high-resolution electrocardiogram in patients with early repolarization pat-

[18] Lin CY, Chung FP, Lin YJ, et al. Gender differences in patients with arrhythmogenic right ventricular dysplasia/cardiomyopathy: Clinical manifestations, electrophysiological properties, substrate characteristics, and prognosis of radiofrequency catheter ablation. International Journal of Cardiology. 2017;**227**:930-937. DOI: 10.1016/j.ijcard.2016.11.055

[19] Dinov B, Bode K, Koenig S, et al. Signal-averaged electrocardiography as a noninvasive tool for evaluating the outcomes after radiofrequency catheter ablation of ventricular tachycardia in patients with ischemic heart disease: Reassessment of an old tool. Circulation: Arrhythmia and Electrophysiology. 2016;**9**(9). pii: e003673. DOI: 10.1161/

[20] Graceffo MA, O'Rourke RA. Cardiac transplant rejection is associated with a decrease in the high-frequency components of the high-resolution, signal averaged electrocardiogram. American Heart Journal. 1996;**132**(4):820-826. DOI: 10.1016/S0002-8703(96)90317-8

[21] Horenstein MS, Idriss SF, Hamilton RM, et al. Efficacy of signal-averaged electrocardiography in the young orthotopic heart transplant patient to detect allograft rejection.

[22] Mendes VN, Pereira TS, Matos VA. Diagnosis of rejection by analyzing ventricular late potentials in heart transplant patients. Arquivos Brasileiros de Cardiologia.

[23] Rejdak K, Rubaj A, Glowniak A, et al. Analysis of ventricular late potentials in signalaveraged ECG of people with epilepsy. Epilepsia. 2011;**52**(11):2118-2124. DOI: 10.1111/

Pediatric Cardiology. 2006;**27**:589-593. DOI: 10.1007/s00246-005-1155-5

2016;**106**(2):136-144. DOI: 10.5935/abc.20160011

tern. Journal of Arrhythmia. 2014;**30**:446-452. DOI: 10.1016/j.joa.2013.12.002

Electrocardiology. 2013;**46**:92-97. DOI: 10.1016/j.jelectrocard.2012.11.010

Electrophysiology. 2015;**8**:863-873. DOI: 10.1161/CIRCEP.114.002551

2010;**43**:48-53. DOI: 10.1016/j.jelectrocard.2009.06.005

10.1016/j.jelectrocard.2010.05.001

48 Interpreting Cardiac Electrograms - From Skin to Endocardium

CIRCEP.115.003673

j.1528-1167.2011.03270.x


[49] Darbar D, Hardy A, Haines JL, et al. Prolonged signal-averaged P-wave duration as an intermediate phenotype for familial atrial fibrillation. Journal of the American College of Cardiology. 2008;**51**(11):1083-1089. DOI: 10.1016/j.jacc.2007.11.058

[37] Ker J. Thyroxine and cardiac electrophysiology—a forgotten physiological duo? Thyroid

[38] Schippinger W, Buchinger W, Schubert B, et al. Late potentials in high resolution ECG in

[39] Kamenicky P, Redheuil A, Roux C, et al. Cardiac structure and function in Cushing's syndrome: A cardiac magnetic resonance imaging study. Journal of Clinical Endocrinology

[40] Lin CC. Analysis of unpredictable intra-QRS potentials in signal-averaged electrocardiograms using an autoregressive moving average prediction model. Medical Engineering

[41] Lin CC. Analysis of abnormal intra-QRS potentials in signal-averaged electrocardiograms using a radial basis function neural network. Sensors. 2016;**16**(10):1580. DOI:

[42] Barbosa PRB, de Souza Bomfim A, Barbosa EC, et al. Spectral turbulence analysis of the signal-averaged electrocardiogram of the atrial activation as predictor of recurrence of idiopathic and persistent atrial fibrillation. International Journal of Cardiology.

[43] Buzea CA, Dan AR, Delcea C, et al. P wave signal-averaged electrocardiography in patients with chronic obstructive pulmonary disease. Romanian Journal of Internal Medicine.

[44] Yogodawa K, Seino Y, Ohara T, et al. Prediction of atrial fibrillation after ischemic stroke using P-wave signal averaged electrocardiography. Journal of Cardiology. 2013;**61**:49-

[45] Aytemir K, Amasyali B, Abali G, et al. The signal-averaged P-wave duration is longer in hypertensive patients with history of paroxysmal atrial fibrillation as compared to those without. International Journal of Cardiology. 2005;**103**:37-40. DOI: 10.1016/j.

[46] Budeus M, Hennersdorf M, Röhlen S, et al. Prediction of atrial fibrillation after coronary artery bypass grafting: The role of chemoreflex sensitivity and P wave signal averaged ECG. International Journal of Cardiology. 2006;**106**:67-74. DOI: 10.1016/j.

[47] Thilen U, Carlson J, Platonov PG, et al. Atrial myocardial pathoelectrophysiology in adults with a secundum atrial septal defect is unaffected by closure of the defect. A study using high resolution signal-averaged orthogonal P-wave technique. International

[48] Auriti A, Aspromonte N, Ceci V, et al. Signal-averaged P-wave in hypertension for the risk of paroxysmal atrial fibrillation. Pacing and Clinical Electrophysiology. 1995;**18**(Part

Journal of Cardiology. 2009;**132**:364-368. DOI: 10.1016/j.ijcard.2007.11.101

thyroid gland dysfunction. Acta Medica Austriaca. 1995;**22**(4):73-74

and Metabolism. 2014;**99**(11):E2144-E2153. DOI: 10.1210/jc.2014-1783

& Physics. 2010;**32**(2):136-144. DOI: 10.1016/j.medengphy.2009.11.001

Research. 2012;**5**(1):8. DOI: 10.1186/1756-6614-5-8

50 Interpreting Cardiac Electrograms - From Skin to Endocardium

2006;**107**:307-316. DOI: 10.1016/j.ijcard.2005.03.073

2015;**53**(4):315-320. DOI: 10.1515/rjim-2015-0040

52. DOI: 10.1016/j.jjcc.2012.08.013

10.3390/s16101580

ijcard.2004.08.027

ijcard.2004.12.062

II):1096


#### **Non-invasive Detection and Compression of Fetal Electrocardiogram** Non-invasive Detection and Compression of Fetal Electrocardiogram

DOI: 10.5772/intechopen.69920

## Xin Gao

Additional information is available at the end of the chapter Xin Gao

http://dx.doi.org/10.5772/intechopen.69920 Additional information is available at the end of the chapter

#### Abstract

Noninvasive detection of fetal electrocardiogram (FECG) from abdominal ECG recordings is highly dependent on typical statistical signal processing techniques such as independent component analysis (ICA), adaptive noise filtering, and multichannel blind deconvolution. In contrast to the previous multichannel FECG extraction methods, several recent schemes for single-channel FECG extraction such as the extended Kalman filter (EKF), extended Kalman smoother (EKS), template subtraction (TS), and support vector regression (SVR) for detecting R waves on ECG, are evaluated via the quantitative metrics such as sensitivity (SE), positive predictive value (PPV), F-score, detection error rate (DER), and range of accuracy. A correlation predictor that combines with multivariable gray model (GM) is also proposed for sequential ECG data compression, which displays better percent root mean-square difference (PRD) than those of Sabah's scheme for fixed and predicted compression ratio (CR). Automatic calculation on fetal heart rate (FHR) on the reconstructed FECG from mixed signals of abdominal ECG recordings is also experimented with sample synthetic ECG data. Sample data on FHR and T/QRS for both physiological case and pathological case are simulated in a 10-min time sequence.

Keywords: noninvasive detection, FECG, FHR, gray prediction, data compression

#### 1. Introduction

Fetal electrocardiogram (FECG) and fetal heart rate (FHR) represent crucial indices for clinical examination and medical diagnosis during pregnancy [1–7, 9–11, 20, 31–36]. In the past decades, multiple systems dynamically monitoring FECG [5, 6, 15, 19, 20, 25–27, 29–31, 35] had been designed for the use of prenatal diagnosis in fetal heart disease, real-time surveillance during both natural and cesarean delivery, as well as the antenatal and intrapartum assessment. Due to the large amount of FECG data for processing in successive monitoring time, enormous storage equipment with durable maintenance is necessary in the design of practical devices [8]: for instance, the double-channel Holter system requires a memory of 82 megabits

© The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons © 2017 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and eproduction in any medium, provided the original work is properly cited.

for sampled data storage with the resolution of 11 bits and 360 Hz for sampling rate per channel every day. Hence, the design of dynamic system urges solutions for better improvements in practical use for noninvasive FECG detection and compression in portable devices and sensing utilities. A variety of typical FECG extraction techniques [2–5, 9, 14–17, 19, 20, 23–26, 35, 36] had been established for both theoretical study and subsequent practical hardware design [18, 25, 29]. A few classical compression methods [13] introduced for efficient data restoration include polynomial fitting, predictive coding, and orthogonal transform-domain compression, where the principle of data compression is to minimize redundancy at comparatively low penalty of distortion and losing useful information [8]. The correlative models exploiting the correlation information between adjacent QRS waves for sequential prediction suggest an efficient scheme for FECG data compression [8].

The classical schemes for noninvasive FECG extraction over the past 30 years mainly comprise of adaptive signal processing with noise cancellation, spatial filtering techniques, and singularvalue decomposition (SVD), to name a few [19, 25]; while the major shortcomings of these schemes were high sensitivity of fetal location and maternal movements, difficulties in extracting P/T waves, and incomplete capture of ECG diagrams [19, 35, 36]. In statistical signal processing, independent component analysis (ICA) [10, 16] aims at computationally separating a mixed signal (with multivariate components) into non-Gaussian signals, where the decomposed signals are assumed to be statistically independent within each other. A variety of methods have been developed for noninvasive FECG extraction since the ICA technique was applied in this research field such as the fourth-order cumulant-based scheme with diagonal approximation proposed by Lathauwer et al. [16], the Joint Approximate Diagonalization of Eigen-matrices (JADE) scheme by Zarzoso [34], Hyvarinen's fast invariant-point method with the ICA principle [10], and the wavelet transform-based infomax algorithm by Jafari and Chambers [12]. Theoretical study on noninvasive FECG extraction methods also employed the ICA-based JADE method with high-order blind identification, the joint detection schemes such as the JADE algorithm with multiple unknown signal extraction, multichannel blind deconvolution [37], and applying the sparse representation of FECG components derived from ICA in the compressed domain [21]. While some previous techniques for noninvasive FECG detection had been considerably mature enough, the challenging issues [4, 28] that have been recognized consist of saving computational cost in abdominal ECG recordings, performing efficient restoration on ECG data, and realizing the practical design (as oriented for low cost, low power, and high integration [29]) for portable FECG monitoring systems. As a result, meeting the balance of recent technical advances with the experimental design on practical systems for noninvasive FECG data-processing devices becomes a crucial task within our investigation.

In this chapter, we present a general study on several categories of algorithms in the field of noninvasive FECG detection, and carry out a performance analysis via several metrics on the state-of-the-art schemes for extracting FECG using sample databases [2, 19, 21]. We proposed a unified approach for the dynamic system design on FECG detection, with a block diagram on noninvasive ECG extraction by collaborating data-processing techniques on weak signal detection and parameter estimation [8]. Utilizing the correlations between adjacent QRS waves of mixed FECG and maternal ECG (MECG), we derived an improved scheme for ECG compression by predicting minimum mean-square error (MMSE), performing integer wavelet transform, quantization, run-length coding, and arithmetic coding for better realization of FECG data compression [8]. Considerably high compression ratio (CR) with feasible lower distortion in contrast to Sabah's scheme is achieved in condition of preserving the most useful message in the compressed FECG data sequence. Simulations rely on the GM(1, 1) model for gray prediction on CR and percent root mean-square difference (PRD) [8]. We also use the sample synthetic ECG data to fulfill the task of automatic estimation on fetal heart rate (FHR) from the reconstructed FECG.

## 2. Methodology

for sampled data storage with the resolution of 11 bits and 360 Hz for sampling rate per channel every day. Hence, the design of dynamic system urges solutions for better improvements in practical use for noninvasive FECG detection and compression in portable devices and sensing utilities. A variety of typical FECG extraction techniques [2–5, 9, 14–17, 19, 20, 23–26, 35, 36] had been established for both theoretical study and subsequent practical hardware design [18, 25, 29]. A few classical compression methods [13] introduced for efficient data restoration include polynomial fitting, predictive coding, and orthogonal transform-domain compression, where the principle of data compression is to minimize redundancy at comparatively low penalty of distortion and losing useful information [8]. The correlative models exploiting the correlation information between adjacent QRS waves for sequential prediction

The classical schemes for noninvasive FECG extraction over the past 30 years mainly comprise of adaptive signal processing with noise cancellation, spatial filtering techniques, and singularvalue decomposition (SVD), to name a few [19, 25]; while the major shortcomings of these schemes were high sensitivity of fetal location and maternal movements, difficulties in extracting P/T waves, and incomplete capture of ECG diagrams [19, 35, 36]. In statistical signal processing, independent component analysis (ICA) [10, 16] aims at computationally separating a mixed signal (with multivariate components) into non-Gaussian signals, where the decomposed signals are assumed to be statistically independent within each other. A variety of methods have been developed for noninvasive FECG extraction since the ICA technique was applied in this research field such as the fourth-order cumulant-based scheme with diagonal approximation proposed by Lathauwer et al. [16], the Joint Approximate Diagonalization of Eigen-matrices (JADE) scheme by Zarzoso [34], Hyvarinen's fast invariant-point method with the ICA principle [10], and the wavelet transform-based infomax algorithm by Jafari and Chambers [12]. Theoretical study on noninvasive FECG extraction methods also employed the ICA-based JADE method with high-order blind identification, the joint detection schemes such as the JADE algorithm with multiple unknown signal extraction, multichannel blind deconvolution [37], and applying the sparse representation of FECG components derived from ICA in the compressed domain [21]. While some previous techniques for noninvasive FECG detection had been considerably mature enough, the challenging issues [4, 28] that have been recognized consist of saving computational cost in abdominal ECG recordings, performing efficient restoration on ECG data, and realizing the practical design (as oriented for low cost, low power, and high integration [29]) for portable FECG monitoring systems. As a result, meeting the balance of recent technical advances with the experimental design on practical systems for noninvasive FECG data-processing devices becomes a crucial task within

In this chapter, we present a general study on several categories of algorithms in the field of noninvasive FECG detection, and carry out a performance analysis via several metrics on the state-of-the-art schemes for extracting FECG using sample databases [2, 19, 21]. We proposed a unified approach for the dynamic system design on FECG detection, with a block diagram on noninvasive ECG extraction by collaborating data-processing techniques on weak signal detection

suggest an efficient scheme for FECG data compression [8].

54 Interpreting Cardiac Electrograms - From Skin to Endocardium

our investigation.

The waveform of ECG as depicted in Figure 1 comprises P, T waves, and the central QRS interval in a regular period of time [8, 24]. Since continuous ECG monitoring explicitly indicates the exposure of regular heart rate, heartbeat rate with amplitude and duration, prior information on the symptoms of potential heart disease is the most important data reference on medical diagnosis. FECG represents weak signals containing a few strong interferences such as MECG with baseline wander, power line interference and additive noise, while the

Figure 1. A diagram on the waveform of ECG in a normal period.

noninvasive techniques aim to eliminate these strong disturbances by directly or indirectly measuring FECG via a few properly located electrodes on the maternal abdomen during pregnancy.

Previous schemes such as fetal scalp electrode monitoring, belong to the category of invasive FECG detection (by either scalp electrode or vaginal ultrasound). However, the invasive schemes have obvious shortcomings such as causing pains and injury to the maternal body, and inducing potential risks on uterus infection to the developing fetus. The fetal ECG detection schemes discussed in this chapter belong to noninvasive techniques, indicating no damage or penetration through maternal or fetal skins.

In general cases on noninvasive detection, the mixed ECG was acquired by multiple electrodes in different locations from both thoracic and abdominal regions on a pregnant woman. For instance, the common diagnostic tool for noninvasive ECG recordings usually adopts 8-lead or 12-lead electrode placement (with symmetric electrodes) [2, 20, 41], which had been derived via clinical validation in a couple of periods. The FECG components in multi-lead abdominal recordings are mutually dependent with each other on the fetal position and the electrical conduction toward the maternal abdominal skin. Due to the variations of each component, calculating linear combinations of multichannel outputs generally enhance the signal-to-noise ratio (SNR) of FECG [20]. Meanwhile, since the main electrical axis of the fetal heart position is a priori uncertain, in order to increase the possibilities that the calculated nonphysiological leads contain significant FECG components, it is often chosen to compute a set of four linear combinations for equal weights, that is, the position angles correspond with 0, 45, 90, and 135 for the 8-lead placement [2, 20]. Similarly, a 12-lead ECG placement (12 leads calculated using 10 electrodes, in which 6 chest electrodes provide information on the heart's horizontal plane and 4 limb electrodes on the heart's vertical plane) illustrates a more cohesive diagram on the accurate electrical activity of the heart by recording information through 12 different perspectives, where the instruction with specific details on the 12-lead placement guide was illustrated in Ref. [41]. Figure 2(a) and (b) illustrate the typical 8-lead electrode placement [2, 20] for noninvasive FECG detection and 12-lead electrode replacement [41] for ECG monitoring, respectively.

Among the representative FECG extraction schemes as mentioned above, while blind source separation (BSS) through ICA [16] was previously regarded as having achieved considerably satisfactory results, the demands for multiple signal inputs (typically 8 channels), in addition to the pre-assumption of linearity between MECG and maternal component in the abdominal ECG recordings and monitoring, were put forward as setbacks toward real-time needs in practical implementations [19]. While simulations on the actual relationship between MECG and maternal component in the abdominal ECG were recorded as single-channel inputs, a few other nonlinear schemes proposed recently can be enumerated as given below: the Bayesian filtering framework using the modified dynamic models via several model-based filters such as the extended Kalman filter (EKF), extended Kalman smoother (EKS), unscented Kalman filter (UKF), and wavelet denoising for synthetic ECG data [22, 26]; the ANFIS system [4], in contrast to normalized least mean squares and polynomial networks, for the identification and extraction of FECG from the aligned MECG; the cascaded framework of EKF (for MECG estimation) with ANFIS (for FECG extraction) on both synthetic and actual ECG data in Non-invasive Detection and Compression of Fetal Electrocardiogram http://dx.doi.org/10.5772/intechopen.69920 57

Figure 2. The illustrations of electrodes on: (a) configuration of 8-lead placement for fetal ECG detection; (b) configuration of 12-lead placement for ECG monitoring: the 6 chest electrodes V1–V6 show the locations on precordial placements, the 4 limb electrodes show the locations on extremity placements (RA–right arm, LA–left arm, RL–right leg, LL–left leg).

contrast to single EKF, EKS [3], template adaptation (TA) [2], nonparametric detection scheme, and modified template subtraction on sequential data processing [19]; the singular spectrum analysis-based fetal heart signal extraction [9], the fetal heartbeat detection algorithm by the integration of Hilbert transform and nonlinear state-space projections [36], and by supporting vector regression [35]; and a few other clinically adopted noninvasive FECG detection methods from multilead abdominal ECG recordings, see Refs. [4, 5, 15, 17, 20, 33], and the references therein.

For nonstationary signals such as ECG, classical evaluation criteria such as the MMSE principle and predictive coding may generate considerably large prediction errors especially when the amplitude of signal depicts quick fluctuation [8]. Consider two adjacent QRS waveforms with strong relativity in successive phases, let x(n) and y(n) be the input and output of ECG signals, respectively; we observe a sequential data of p points of the former QRS waveform in order to predict the present waveform at the minor cost of generating prediction errors. The predictor output is expressed as [8]:

$$\mathbf{y}(n) = \sum\_{i=1}^{p} \alpha\_i \mathbf{x}(n - \mathbf{i} - T) \tag{1}$$

where α<sup>i</sup> denotes the coefficients of system cascades which can be obtained by Yule-Walker equations and T stands for the average time period between the intervals of R waves.

The prediction error ε(n) can be calculated via [8]

noninvasive techniques aim to eliminate these strong disturbances by directly or indirectly measuring FECG via a few properly located electrodes on the maternal abdomen during

Previous schemes such as fetal scalp electrode monitoring, belong to the category of invasive FECG detection (by either scalp electrode or vaginal ultrasound). However, the invasive schemes have obvious shortcomings such as causing pains and injury to the maternal body, and inducing potential risks on uterus infection to the developing fetus. The fetal ECG detection schemes discussed in this chapter belong to noninvasive techniques, indicating no damage

In general cases on noninvasive detection, the mixed ECG was acquired by multiple electrodes in different locations from both thoracic and abdominal regions on a pregnant woman. For instance, the common diagnostic tool for noninvasive ECG recordings usually adopts 8-lead or 12-lead electrode placement (with symmetric electrodes) [2, 20, 41], which had been derived via clinical validation in a couple of periods. The FECG components in multi-lead abdominal recordings are mutually dependent with each other on the fetal position and the electrical conduction toward the maternal abdominal skin. Due to the variations of each component, calculating linear combinations of multichannel outputs generally enhance the signal-to-noise ratio (SNR) of FECG [20]. Meanwhile, since the main electrical axis of the fetal heart position is a priori uncertain, in order to increase the possibilities that the calculated nonphysiological leads contain significant FECG components, it is often chosen to compute a set of four linear combinations for equal weights, that is, the position angles correspond with 0, 45, 90, and 135 for the 8-lead placement [2, 20]. Similarly, a 12-lead ECG placement (12 leads calculated using 10 electrodes, in which 6 chest electrodes provide information on the heart's horizontal plane and 4 limb electrodes on the heart's vertical plane) illustrates a more cohesive diagram on the accurate electrical activity of the heart by recording information through 12 different perspectives, where the instruction with specific details on the 12-lead placement guide was illustrated in Ref. [41]. Figure 2(a) and (b) illustrate the typical 8-lead electrode placement [2, 20] for noninvasive FECG

detection and 12-lead electrode replacement [41] for ECG monitoring, respectively.

Among the representative FECG extraction schemes as mentioned above, while blind source separation (BSS) through ICA [16] was previously regarded as having achieved considerably satisfactory results, the demands for multiple signal inputs (typically 8 channels), in addition to the pre-assumption of linearity between MECG and maternal component in the abdominal ECG recordings and monitoring, were put forward as setbacks toward real-time needs in practical implementations [19]. While simulations on the actual relationship between MECG and maternal component in the abdominal ECG were recorded as single-channel inputs, a few other nonlinear schemes proposed recently can be enumerated as given below: the Bayesian filtering framework using the modified dynamic models via several model-based filters such as the extended Kalman filter (EKF), extended Kalman smoother (EKS), unscented Kalman filter (UKF), and wavelet denoising for synthetic ECG data [22, 26]; the ANFIS system [4], in contrast to normalized least mean squares and polynomial networks, for the identification and extraction of FECG from the aligned MECG; the cascaded framework of EKF (for MECG estimation) with ANFIS (for FECG extraction) on both synthetic and actual ECG data in

pregnancy.

or penetration through maternal or fetal skins.

56 Interpreting Cardiac Electrograms - From Skin to Endocardium

$$\mathbf{x}(n) = \mathbf{x}(n) - \mathbf{y}(n) = \mathbf{x}(n) - \sum\_{i=1}^{p} \alpha\_i \mathbf{x}(n - i - T) \tag{2}$$

where a set of consecutive p points represents the orders of correlation predictor.

Since the prediction is processed between two adjacent QRS waveforms, let us denote such kind of prediction as the twin-R correlative prediction [8]. In mean-square scales, we express the energy Ep of prediction errors as [8]

$$\begin{aligned} E\_p &= E[\varepsilon^2(n)] = E\left\{ \left[ \mathbf{x}(n) - \sum\_{i=1}^p a\_i \mathbf{x}(n-i-T) \right]^2 \right\} \\ &= E\left\{ \mathbf{x}^2(n) - 2\sum\_{i=1}^p a\_i \mathbf{x}(n)\mathbf{x}(n-i-T) + \sum\_{i=1}^p a\_i \sum\_{j=1}^p a\_j \mathbf{x}(n-i-T)\mathbf{x}(n-j-T) \right\} \\ &= R(0) - 2\sum\_{i=1}^p a\_i R(i+T) + \sum\_{i=1}^p a\_i \sum\_{j=1}^p a\_j R(j-i) \end{aligned} \tag{3}$$

The correlation coefficients of input ECG waves can be calculated via [8]

$$R(m) = \frac{1}{N} \sum\_{n=1}^{N-1} x(n)x(n-m) \tag{4}$$

Simplifying Eq. (4) by taking <sup>∂</sup>Ep <sup>∂</sup>α<sup>i</sup> ¼ 0 to obtain a minimum for Ep (m ¼ 0, 1,…, p �1) yields [8]

$$\sum\_{i=1}^{p-1} \alpha\_i \mathbb{R}(m-i) = \mathbb{R}(m+T), \qquad i = 0, 1, \ldots, p-1 \tag{5}$$

Constructing the matrix of correlation coefficients by combining Eqs. (3)–(5) yields the linear algebraic equations as follows [8]

$$
\begin{bmatrix} R(0) & R(1) & \cdots & R(p-1) \\ R(1) & R(0) & \cdots & R(p-2) \\ \vdots & \vdots & \ddots & \vdots \\ R(p-1) & R(p-2) & \cdots & R(0) \end{bmatrix} \begin{bmatrix} a\_0 \\ a\_1 \\ \vdots \\ a\_{p-1} \end{bmatrix} = \begin{bmatrix} R(T) \\ R(T+1) \\ \vdots \\ R(T+p-1) \end{bmatrix} \tag{6}
$$

Solving the equation group in Eq. (6) as above yields the numerical coefficients of each αi.

The lifting wavelet transform (LWT) has been recognized as a strong implementation when combined with a few algorithms such as integer square zero-tree wavelet coding [8]. Splitting, predicting, and updating symbols, are three steps in the lifting scheme of a typical LWT. The proposed scheme is presented as follows: the first step is to split the ECG data sequence {ej} into two sequences {oj-1} and {ej-1} that stands for odd and even numerals via [8]

$$\text{split}\left(e\_{\hat{\jmath}}\right) = \left(e\_{\hat{\jmath}-1}, o\_{\hat{\jmath}-1}\right) \tag{7}$$

Second, with respect to the predictor filter group P and the earlier even sequence {ej-1}, the odd sequence {oj-1} is predicted by exploiting correlativity information such as [8]

$$
\sigma\_{j-1} := \sigma\_{j-1} - P(\mathfrak{e}\_{j-1}) \tag{8}
$$

The last step of updating claims that some integral characteristics as those of integrity for the original {ej} need to be preserved for constructing a better subset {ej-1}. As a result, we adopt an updating filter U that exploits the discrepancy between a specific parameter (i.e., mean, variance, or wavelet vanishing moments) and {ej}, where this step is proceeded by [8]

$$e\_{j-1} := e\_{j-1} + \mathcal{U}(o\_{j-1}) \tag{9}$$

The inverse transform of LWT for signal reconstruction can be similarly expressed via [8]

$$\begin{cases} e\_{j-1} := e\_{j-1} - \mathcal{U}(o\_{j-1}) \\ o\_{j-1} := o\_{j-1} + P(e\_{j-1}) \\ e\_j = \text{Merge}\left(e\_{j-1}, o\_{j-1}\right) \end{cases} \tag{10}$$

The iteration procedures as performed in Ref. [8], applied a (4, 2) LWT for the decomposition and reconstruction of ECG signals, which can be proceeded by [8]

$$\begin{cases} o\_{\boldsymbol{\gamma}}[n] := e\_{\boldsymbol{\gamma}-1}[n] + \left\lfloor \frac{1}{16} \left\{ (e\_{\boldsymbol{\gamma}-1}[n+2]) - 9(e\_{\boldsymbol{\gamma}-1}[n+2]) + e\_{\boldsymbol{\gamma}-1}[n-1]) + e\_{\boldsymbol{\gamma}-1}[n-1] \right\} + \frac{1}{2} \right\rfloor \\\\ e\_{\boldsymbol{\gamma}}[n] := e\_{\boldsymbol{\gamma}-1}[n] + \left\lfloor \frac{1}{4} (o\_{\boldsymbol{\gamma}}[n] + o\_{\boldsymbol{\gamma}}[n-1]) + \frac{1}{2} \right\rfloor \end{cases} \tag{11}$$

where ⌊:⌋ denotes the execution of the round-off operation.

Since the prediction is processed between two adjacent QRS waveforms, let us denote such kind of prediction as the twin-R correlative prediction [8]. In mean-square scales, we express

αixðn � i � TÞ

9 = ;

j¼1

<sup>∂</sup>α<sup>i</sup> ¼ 0 to obtain a minimum for Ep (m ¼ 0, 1,…, p �1) yields [8]

αiRðm � iÞ ¼ Rðm þ TÞ, i ¼ 0, 1,…, p–1 (5)

αjxðn � i � TÞxðn � j � TÞ

xðnÞxðn � mÞ (4)

RðTÞ RðT þ 1Þ ⋮ RðT þ p � 1Þ

9 = ;

(3)

(6)

p

i¼1 αi X p

αjRðj � iÞ

the energy Ep of prediction errors as [8]

58 Interpreting Cardiac Electrograms - From Skin to Endocardium

ðnÞ � 2

X p

i¼1

X p�1

i¼1

<sup>ð</sup>nÞ� ¼ E xðnÞ �<sup>X</sup>

X p

i¼1

<sup>α</sup>iRð<sup>i</sup> <sup>þ</sup> <sup>T</sup>Þ þ<sup>X</sup>

< : p

" #<sup>2</sup> 8

<sup>α</sup>ixðnÞxð<sup>n</sup> � <sup>i</sup> � <sup>T</sup>Þ þ<sup>X</sup>

p

i¼1 αi X p

The correlation coefficients of input ECG waves can be calculated via [8]

Rð0Þ Rð1Þ ⋯ Rðp � 1Þ Rð1Þ Rð0Þ ⋯ Rðp � 2Þ ⋮ ⋮⋱⋮ Rðp � 1Þ Rðp � 2Þ ⋯ Rð0Þ

<sup>R</sup>ðmÞ ¼ <sup>1</sup> N N X�1 n¼1

j¼1

Constructing the matrix of correlation coefficients by combining Eqs. (3)–(5) yields the linear

Solving the equation group in Eq. (6) as above yields the numerical coefficients of each αi.

into two sequences {oj-1} and {ej-1} that stands for odd and even numerals via [8]

sequence {oj-1} is predicted by exploiting correlativity information such as [8]

The lifting wavelet transform (LWT) has been recognized as a strong implementation when combined with a few algorithms such as integer square zero-tree wavelet coding [8]. Splitting, predicting, and updating symbols, are three steps in the lifting scheme of a typical LWT. The proposed scheme is presented as follows: the first step is to split the ECG data sequence {ej}

Second, with respect to the predictor filter group P and the earlier even sequence {ej-1}, the odd

α0 α1 ⋮ α<sup>p</sup>�<sup>1</sup>

split ðejÞ¼ðej�<sup>1</sup>, oj�<sup>1</sup>Þ (7)

i¼1

Ep <sup>¼</sup> <sup>E</sup>½ε<sup>2</sup>

<sup>¼</sup> E x<sup>2</sup>

8 < :

¼ Rð0Þ � 2

Simplifying Eq. (4) by taking <sup>∂</sup>Ep

algebraic equations as follows [8]

The advantages of LWT compared to other wavelet transform methods are displayed in the following scenarios [8]: (i) less dependence for the down-sampling of low pass and high pass signal components and easier realization on the inverse operation of LWT; (ii) reduced execution times by avoiding calculating floating points coming from the integer coefficients; (iii) the implementation of hardware circuits is also much easier; and (iv) guaranteed quality for signal recovery free of boundary continuation in any type.

The procedure of our proposed twin-R correlation predictor for improving sequential ECG compression is presented as below [8]: let us denote the implement D as the first-order time delay and Pi as the location of the ith R-wave; A<sup>j</sup> ¼ {αj,0, αj,1, …, αj,p-1} stands for the aggregated coefficients for the twin-R interval of the jth ECG sequence. We take the following steps to perform this task:

Step 1. For the original ECG signal with length N, initially perform the first-order prediction to reduce the DC components of the signals; let z(n) be the redundancy within smooth district of the ECG samples calling for elimination, the residual term z(n) is now expressed as

$$z(n) = \mathbf{x}(n) - \mathbf{x}(n-1), \qquad n = \text{ 0, 1, } \ldots, N-1. \tag{12}$$

Step 2. While Pi, the locations of R-wave for each QRS waveform, have been identified, compute each Ti by deducing Ti ¼ Piþ<sup>1</sup> � Pi, and estimate the central position of the adjacent twin-R waves, where mi ¼ (Piþ<sup>1</sup> þ Pi )/2. This step speeds up higher recognition rate and operation time, and bears negative effects such as noise interference or baseline shift.

Step 3. Perform the correlation prediction for z(n) similar to Eq. (2):

$$d(n) = z(n) - \sum\_{k=0}^{p-1} \alpha\_{\bar{i}} z(n - i - T\_{i-1}),\\ m\_i - l \le m\_i + l,\\ l = \min(m\_{i-1} - m\_{i-2}, m\_i - m\_{i-1})/2,\\ i = 1, \dots, N\_{\bar{i}}.\tag{13}$$

where d(n) denotes the signal of prediction error, p and Nj stand for the order of predictor and the R-wave counts of the jth ECG data sequence, respectively. Without loss of generality, we adopt p ¼ 4. The kth prediction coefficient αj,k of each compressed 16-bit ECG sequence was obtained via the splitting process in Step 1. Due to the slow drift for the QRS waveform, an interval of 30 seconds was used to partition this data stream. Note that we implemented the same predictors for continuous QRS waveforms of the same ECG data in order to reduce computational cost and enhance the efficiency for ECG data compression.

Step 4. Update the (4, 2) LWT on the signal d(n) via Eq. (12), where the length of w(n) is preserved as N. For the subband signal w(n) containing {oi(n) ¼ 1,2,3,4} and the approximated signal e4(n), their length are constructed as N/2, N/4, N/8, N/16, and N/16, respectively.

Step 5. Perform scalar quantization, run-length coding, and arithmetic coding for w(n). While a few zero-coefficients appear after quantization toward w(n), these successive zeros can be removed via run-length searching so as to shorten the ECG data sequence. Variable quantization coefficients are selected in this procedure; after run-length coding each ECG data sequence is merged by three parts: the constructed bit streams, vector Pi for R-wave localization, and the twin-R predictor coefficients A<sup>j</sup> ¼ {αj,0, αj,1, …, αj,p-1}.

The flowchart of this scheme as described above is depicted in the block diagram of Figure 3, where the prediction step is associated with the compressed ECG data stream.

Since the proposed scheme is invertible for decompression, we need to observe the correlativity and fluctuation tendency between two sequential ECG data; hence, a Lemma is presented for the derivation of ECG data prediction via the single-variable gray model [8].

Figure 3. The block diagram of unified twin-R predictive method for ECG sequential data compression.

Lemma 1 [8]: Consider a stationary sequence T<sup>0</sup> ¼ {T0(k)| k ¼ 1,…, n} ¼ {T0(1),…, T0(n)}, where k represents the time point. Let us observe a number of m sequences as reference where Ti ¼ {Ti(k) | k ¼ 1, 2,…, n} ¼ {Ti(1), Ti(2),…, Ti(n)}, i ¼ 1, 2,…, n. Define ξ<sup>k</sup> as the correlation coefficient of the kth reference sequence with respect to the starting sequence T<sup>0</sup> at time k,

waves, where mi ¼ (Piþ<sup>1</sup> þ Pi

60 Interpreting Cardiac Electrograms - From Skin to Endocardium

p�1

k¼0

<sup>d</sup>ðnÞ ¼ <sup>z</sup>ðnÞ �<sup>X</sup>

and bears negative effects such as noise interference or baseline shift. Step 3. Perform the correlation prediction for z(n) similar to Eq. (2):

computational cost and enhance the efficiency for ECG data compression.

twin-R predictor coefficients A<sup>j</sup> ¼ {αj,0, αj,1, …, αj,p-1}.

)/2. This step speeds up higher recognition rate and operation time,

(13)

αj, kzðn � i � Ti�<sup>1</sup>Þ, mi � l ≤ mi þ l, l ¼ minð Þ mi�<sup>1</sup> � mi�<sup>2</sup>, mi � mi�<sup>1</sup> =2, i ¼ 1,…, Nj:

where d(n) denotes the signal of prediction error, p and Nj stand for the order of predictor and the R-wave counts of the jth ECG data sequence, respectively. Without loss of generality, we adopt p ¼ 4. The kth prediction coefficient αj,k of each compressed 16-bit ECG sequence was obtained via the splitting process in Step 1. Due to the slow drift for the QRS waveform, an interval of 30 seconds was used to partition this data stream. Note that we implemented the same predictors for continuous QRS waveforms of the same ECG data in order to reduce

Step 4. Update the (4, 2) LWT on the signal d(n) via Eq. (12), where the length of w(n) is preserved as N. For the subband signal w(n) containing {oi(n) ¼ 1,2,3,4} and the approximated

Step 5. Perform scalar quantization, run-length coding, and arithmetic coding for w(n). While a few zero-coefficients appear after quantization toward w(n), these successive zeros can be removed via run-length searching so as to shorten the ECG data sequence. Variable quantization coefficients are selected in this procedure; after run-length coding each ECG data sequence is merged by three parts: the constructed bit streams, vector Pi for R-wave localization, and the

The flowchart of this scheme as described above is depicted in the block diagram of Figure 3,

Since the proposed scheme is invertible for decompression, we need to observe the correlativity and fluctuation tendency between two sequential ECG data; hence, a Lemma is presented for the

where the prediction step is associated with the compressed ECG data stream.

Figure 3. The block diagram of unified twin-R predictive method for ECG sequential data compression.

derivation of ECG data prediction via the single-variable gray model [8].

signal e4(n), their length are constructed as N/2, N/4, N/8, N/16, and N/16, respectively.

$$\xi\_k = \frac{\min\_i \min\_k \left| T\_0(k) - T\_i(k) \right| + \rho \max\_i \max\_k \left| T\_0(k) - T\_i(k) \right|}{\left| T\_0(k) - T\_i(k) \right| + \rho \max\_i \max\_k \left| T\_0(k) - T\_i(k) \right|} \tag{14}$$

where ρ ∈ [0,1) denotes the resolution coefficient (and without loss of generality it is often taking expectation of ρ ¼ 0.5), min<sup>i</sup> minkjT0ðkÞ � TiðkÞj and max<sup>i</sup> max<sup>i</sup> jTiðkÞ � T0ðkÞj represents the minimum and maximum difference value between two-levels, respectively. In the gray system theory, ri <sup>¼</sup> <sup>1</sup> n P<sup>n</sup> <sup>k</sup>¼<sup>1</sup> <sup>ξ</sup><sup>k</sup> denotes the relevance of sequence Ti to <sup>T</sup>0. Geometric similarity on two sequences reflects the degree of correlativity.

Consider the ith sequence Ti¼ {Ti(1), Ti(2),…, Ti(n)}, the initialized sequence of original Ti is written as T ¼ ð1, Tð2Þ=Tð1Þ, …TðnÞ=Tð1ÞÞ, and the correlation factor σ<sup>i</sup> can be computed via <sup>σ</sup><sup>i</sup> <sup>¼</sup> <sup>X</sup> N k¼1 kTiðkÞ �<sup>X</sup> N k¼1 TiðkÞ X N k¼1 k n , which has the possibility of being either positive or negative.

For instance, in the simplest case of i ¼ 1, 2, the sequential expression of Ti is formulated as [8]

$$T\_i = \left(1, \frac{T\_i(1)}{T\_i(2)}, \frac{T\_i(1)}{T\_i(3)}, \dots, \frac{T\_i(1)}{T\_i(k)}\right); i = 1, \ 2; k = N \tag{15}$$

According to Lemma 1, the degree of correlativity is measured by solving Eq. (14). Note that if signðσ1=σnÞsignðσ2=σnÞ ¼ 1, a positive relevance is justified between T<sup>1</sup> and T2; conversely, a negative relevance is justified when signðσ1=σnÞsignðσ2=σn޼�1. In more general cases such as ECG data sequence, the correlation factor σ<sup>n</sup> can be approximately estimated via [8]

$$
\sigma\_n = \sum\_{k=1}^n k^2 - \left(\sum\_{k=1}^n k^2\right) \bigg/ n \tag{16}
$$

In the gray system theory, the single variable GM(1, 1) model is often applied to predict the upcoming sequence number and estimate the missed numerical values between time intervals, for the processing of ECG data, we just equalize the corresponding parameters in the time domain, and deduce the gray predictor in the scenario as follows:

The least square (LS) update consists of a whitening procedure through constructing a differential equation in the whitening process with its estimate, and a discretization process for the residuals, which is formulated as [8]:

$$\frac{dT\_i^{(1)}}{dt} + aT\_i^{(1)} = u,\\ \hat{a} = \left(a, u\right)^T \tag{17}$$

$$\hat{\mathbf{a}} = \left(\mathbf{B}\_i^T \mathbf{B}\_i\right)^{-1} \mathbf{B}\_i^T \mathbf{Y}\_1 \tag{18}$$

$$\mathbf{B}\_{i} = \begin{bmatrix} -\frac{1}{2} (T\_{i}^{(1)}(1) + T\_{i}^{(1)}(2)) & 1\\ -\frac{1}{2} (T\_{i}^{(1)}(2) + T\_{i}^{(1)}(3)) & 1\\ \vdots & \vdots\\ -\frac{1}{2} (T\_{i}^{(1)}(n-1) + T\_{i}^{(1)}(n)) & 1 \end{bmatrix}, \mathbf{Y}\_{i} = \begin{bmatrix} T\_{i}^{(1)}(2) \\ T\_{i}^{(1)}(3) \\ \vdots \\ T\_{i}^{(1)}(n) \end{bmatrix} \tag{19}$$

where B<sup>i</sup> and Y<sup>i</sup> denote the data matrix and data vector of GM(1, 1) model, respectively.

A general solution to the matrix equations above is given by [8]

$$T\_i^{(1)}(k+1) = \left(T\_i^{(0)}(1) - \frac{\mu}{a}\right)e^{ak} + \frac{\mu}{a} \tag{20}$$

Determining the model parameters (a, u) yields the past or upcoming numerical values from this predictive GM(1, 1) model. Note that the constructed gray model indicates coincidence with the time-variant extrapolate prediction. In harsh conditions, due to the scarcity of prior information and ambiguity of system on ECG data processing, this predictive GM model is useful since only four adjacent continuous data points are needed from the least data sample.

Because the quality of FECG reflects crucial information on fetal heart rate (FHR) and its beat-tobeat variability [9], the cascaded system design for noninvasive FECG extraction may often involve a post-processing stage such as adaptive noise cancellation or wavelet denoising [12, 22, 34]. FHR is usually estimated via the ratio of 60 to the average time period (s) on a sequence of adjacent intervals from R waves, while estimating FHR technically requires shaping fetal QRS complexes by capturing data via multichannel maternal abdominal ECG recordings [2, 15, 19– 21, 23, 26, 30, 32], and by adopting a few other sensing technologies through the Doppler ultrasound devices [37], fetal phonocardiography [1], as well as superconducting magneto-cardiography [33]. Wearable devices on ECG rhythm recording via potential mapping on the wrist/ arm surface skin [42] also urge collaborative concerns from industry field toward our theoretically proposed algorithmic study.

#### 3. Performance metrics

The diagnostic tests in biomedical engineering often employ a set of performance metrics in order to evaluate the validity of tests in the subjects on study. In ECG detection, the parameters of true positive (TP), false negative (FN), and false positive (FP) are called from the counts of detected R-peaks. We denote TP as the number of correctly detected R peaks, FN stands for the number of missed R peaks, and FP represents the number of noise spikes detected as R peaks. Hence, the measures of sensitivity (SE) and positive predictive values (PPV) are formulated as [19, 30]:

$$SE = \frac{TP}{TP + FN} \times 100\% \tag{21}$$

Non-invasive Detection and Compression of Fetal Electrocardiogram http://dx.doi.org/10.5772/intechopen.69920 63

$$PPV = \frac{TP}{TP + FP} \times 100\% \tag{22}$$

The F-score, known as the harmonic mean of SE and PPV, is expressed as [2, 21]:

B<sup>i</sup> ¼

62 Interpreting Cardiac Electrograms - From Skin to Endocardium

proposed algorithmic study.

3. Performance metrics

� 1 2 ðT<sup>ð</sup>1<sup>Þ</sup>

� 1 2 ðT<sup>ð</sup>1<sup>Þ</sup>

A general solution to the matrix equations above is given by [8]

T<sup>ð</sup>1<sup>Þ</sup>

� 1 2 ðT<sup>ð</sup>1<sup>Þ</sup>

<sup>i</sup> <sup>ð</sup>1Þ þ <sup>T</sup><sup>ð</sup>1<sup>Þ</sup>

<sup>i</sup> <sup>ð</sup>2Þ þ <sup>T</sup><sup>ð</sup>1<sup>Þ</sup>

<sup>i</sup> <sup>ð</sup><sup>n</sup> � <sup>1</sup>Þ þ <sup>T</sup><sup>ð</sup>1<sup>Þ</sup>

<sup>i</sup> <sup>ð</sup><sup>k</sup> <sup>þ</sup> <sup>1</sup>Þ ¼ <sup>T</sup><sup>ð</sup>0<sup>Þ</sup>

where B<sup>i</sup> and Y<sup>i</sup> denote the data matrix and data vector of GM(1, 1) model, respectively.

<sup>i</sup> ð2ÞÞ 1

, Y<sup>i</sup> ¼

T<sup>ð</sup>1<sup>Þ</sup> <sup>i</sup> ð2Þ T<sup>ð</sup>1<sup>Þ</sup> <sup>i</sup> ð3Þ ⋮ T<sup>ð</sup>1<sup>Þ</sup> <sup>i</sup> ðnÞ

<sup>a</sup> (20)

� 100% (21)

(19)

<sup>i</sup> ð3ÞÞ 1 ⋮ ⋮

<sup>i</sup> ðnÞÞ 1

<sup>i</sup> <sup>ð</sup>1Þ � <sup>u</sup> a

e ak <sup>þ</sup> u

� �

Determining the model parameters (a, u) yields the past or upcoming numerical values from this predictive GM(1, 1) model. Note that the constructed gray model indicates coincidence with the time-variant extrapolate prediction. In harsh conditions, due to the scarcity of prior information and ambiguity of system on ECG data processing, this predictive GM model is useful since only four adjacent continuous data points are needed from the least data sample. Because the quality of FECG reflects crucial information on fetal heart rate (FHR) and its beat-tobeat variability [9], the cascaded system design for noninvasive FECG extraction may often involve a post-processing stage such as adaptive noise cancellation or wavelet denoising [12, 22, 34]. FHR is usually estimated via the ratio of 60 to the average time period (s) on a sequence of adjacent intervals from R waves, while estimating FHR technically requires shaping fetal QRS complexes by capturing data via multichannel maternal abdominal ECG recordings [2, 15, 19– 21, 23, 26, 30, 32], and by adopting a few other sensing technologies through the Doppler ultrasound devices [37], fetal phonocardiography [1], as well as superconducting magneto-cardiography [33]. Wearable devices on ECG rhythm recording via potential mapping on the wrist/ arm surface skin [42] also urge collaborative concerns from industry field toward our theoretically

The diagnostic tests in biomedical engineering often employ a set of performance metrics in order to evaluate the validity of tests in the subjects on study. In ECG detection, the parameters of true positive (TP), false negative (FN), and false positive (FP) are called from the counts of detected R-peaks. We denote TP as the number of correctly detected R peaks, FN stands for the number of missed R peaks, and FP represents the number of noise spikes detected as R peaks. Hence, the measures of sensitivity (SE) and positive predictive values (PPV) are formulated as [19, 30]:

SE <sup>¼</sup> TP

TP þ FN

$$F\text{-score} = 2 \cdot \frac{SE \times PPV}{SE + PPV} = \frac{2 \times TP}{2 \times TP + FP + FN} \tag{23}$$

Since the total number of R-wave peaks is the sum of TP, FN, and FP, the detection error rate (DER) is now denoted as [30]:

$$DER = \frac{FP + FN}{TP + FP + FN} \times 100\% \tag{24}$$

For each DER, the metric of accuracy ¼ 1 – DER yields the same expression as defined in Ref. [19].

The percent root mean-square difference (PRD) represents a fidelity measure for some data compression scheme on the reconstructed/predicted signal in contrast to the original ECG, where the PRD value is numerically calculated as follows [13]:

$$PRD = \frac{\sqrt{\sum\_{n=1}^{p} \left[ \mathbf{x}(n) - \mathbf{y}(n) \right]^2}}{\sum\_{n=1}^{p} \mathbf{x}^2(n)} \times 100\tag{25}$$

where x(n) and y(n) correspondingly represent samples of the original and the reconstructed/ predicted ECG data sequences and the length of sequence is p.

Regarding the compression ratio (CR) defined as the proportion of uncompressed size to compressed size for a finite data sequence, or the ratio of uncompressed data rate to compressed data rate for streaming media signals of infinite size such as video or audio [38], for each compression scheme, there is a PRD value corresponding to a required CR.

For synthetic ECG data, consider the abdominal ECG w(n) in case of a single-channel dynamic model, which is nonlinearly synthesized via the MECG m(n), FECG f(n), and the additive white noise η(n), and hence, the composite signal is modeled as [19]:

$$
\hat{\sigma}w(n) = \hat{m}(n) + \hat{f}(n) = \hat{m}(n) + f(n) + \eta(n) \tag{26}
$$

where <sup>m</sup>^ <sup>ð</sup>n<sup>Þ</sup> and ^<sup>f</sup> <sup>ð</sup>n<sup>Þ</sup> denote the nonlinear expressions of MECG and FECG, respectively. Since the noise power in η(n) can be adjusted to test the performance of each noninvasive FECG detection scheme [19, 30], for some ECG data sequence with a length of p periodical R peaks, the fetal to maternal signal-to-noise ratio (fmSNR) can be calculated via [19]:

$$fm\text{SNR} = 10\log\_{10}\left(\frac{\sum\_{n=1}^{p} \left[\hat{f}(n)\right]^2}{\sum\_{n=1}^{p} \left[\hat{m}(n)\right]^2}\right) \tag{27}$$

Up till now, we have presented a concise study for the keynote noninvasive techniques and quantitative metrics on FECG detection, with an emphasis on single-channel FECG extraction via nonlinear dynamic models. We proposed a flowchart of processing ECG data sequence by means of LWT and the unified twin-R correlation predictor by implementing GM(1,1) model for ECG data compression.

In the next section, we will present three sets of experiments for the qualitative and quantitative evaluation on several noninvasive FECG detection schemes [2, 19, 21], the proposed twin-R correlative ECG compression scheme via a widely used ECG database [8], and automatic FHR estimation over a sample sequence of synthetic ECG data [40].

## 4. Experimental results

We employ sample ECG data from several databases to perform our experimental study: the CinC Challenging Data as referenced in Ref. [2] (also known as the Physionet challenge dataset in Ref. [21]), a noninvasive fetal ECG database in Ref. [19], sample ECG sequences from MIT-BIS Arrhythmia Dataset [8], and some synthetic ECG data from Dr. Igal A. Sebag's example [39]. The main set of experiments with demographic data on sample patients with clinical/ synthetic information were summarized in Table 1.

The first set of experiments mainly recorded the quantitative evaluations on several representative noninvasive FECG detection schemes based on single-channel abdominal ECG recordings. We studied the test by Panigrahy and Sahu [19] where the QRS complex of FECG displays the most visible features after the preprocessing step of eliminating baseline wander and power line interference from MECG, then each scheme using noninvasive FECG database was implemented to test the detection performance within 60 s of measuring R waves.

The numerical results for SE, PPV, F-score, and DER on nine methods for FECG detection are illustrated in Table 2, where the first column chronologically enumerated the tested FECG detection schemes which correspond to the average score on each measure for the recorded R waves, and the last column specified the range of accuracy over a certain length of time duration [2, 19, 21].

From Table 2, we justify that the SE metric on eight FECG detection schemes achieved over 90% except the TA scheme; the metrics of PPV and F-score on seven schemes reached over 90% except for SVD and TA; regarding DER, SVD shows the worst performance while it is still as low as 18.7%. Among all the five parameters on the referred quantitative analysis, EKS þ ANFIS displays the best overall scores for each metric, while EKF þ ANFIS indicates the second best results on F-score, DER, and other range of accuracy.


f mSNR ¼ 10 log10

FHR estimation over a sample sequence of synthetic ECG data [40].

synthetic information were summarized in Table 1.

for ECG data compression.

64 Interpreting Cardiac Electrograms - From Skin to Endocardium

4. Experimental results

duration [2, 19, 21].

X p

0

BBBB@

Up till now, we have presented a concise study for the keynote noninvasive techniques and quantitative metrics on FECG detection, with an emphasis on single-channel FECG extraction via nonlinear dynamic models. We proposed a flowchart of processing ECG data sequence by means of LWT and the unified twin-R correlation predictor by implementing GM(1,1) model

In the next section, we will present three sets of experiments for the qualitative and quantitative evaluation on several noninvasive FECG detection schemes [2, 19, 21], the proposed twin-R correlative ECG compression scheme via a widely used ECG database [8], and automatic

We employ sample ECG data from several databases to perform our experimental study: the CinC Challenging Data as referenced in Ref. [2] (also known as the Physionet challenge dataset in Ref. [21]), a noninvasive fetal ECG database in Ref. [19], sample ECG sequences from MIT-BIS Arrhythmia Dataset [8], and some synthetic ECG data from Dr. Igal A. Sebag's example [39]. The main set of experiments with demographic data on sample patients with clinical/

The first set of experiments mainly recorded the quantitative evaluations on several representative noninvasive FECG detection schemes based on single-channel abdominal ECG recordings. We studied the test by Panigrahy and Sahu [19] where the QRS complex of FECG displays the most visible features after the preprocessing step of eliminating baseline wander and power line interference from MECG, then each scheme using noninvasive FECG database

The numerical results for SE, PPV, F-score, and DER on nine methods for FECG detection are illustrated in Table 2, where the first column chronologically enumerated the tested FECG detection schemes which correspond to the average score on each measure for the recorded R waves, and the last column specified the range of accuracy over a certain length of time

From Table 2, we justify that the SE metric on eight FECG detection schemes achieved over 90% except the TA scheme; the metrics of PPV and F-score on seven schemes reached over 90% except for SVD and TA; regarding DER, SVD shows the worst performance while it is still as low as 18.7%. Among all the five parameters on the referred quantitative analysis, EKS þ ANFIS displays the best overall scores for each metric, while EKF þ ANFIS indicates the

second best results on F-score, DER, and other range of accuracy.

was implemented to test the detection performance within 60 s of measuring R waves.

n¼1 ½ ^<sup>f</sup> <sup>ð</sup>nÞ�<sup>2</sup>

X p

<sup>½</sup>m^ <sup>ð</sup>nÞ�<sup>2</sup>

1

CCCCA

(27)

n¼1

Table 1. Summary on the main set of experiments for noninvasive techniques on ECG detection and monitoring.


Table 2. Quantitative scores of average SE, PPV, F-score, DER, and range of accuracy on noninvasive FECG detection schemes using a sample FECG database (duration ¼ 60 s).

The second set of experiments was conducted by using the standard database on MIT-BIS Arrhythmia [8] with some original sample data of mixed MECG and FECG. We first investigate the predicted output of errors in comparison to real data from a mixed ECG sequence. Figure 4 displays four subplots, where (a) depicts an original sequence in time interval [0:1600]

Figure 4. (a) Originally detected FECG; sequential data output by: (b) single linear prediction; (c) fourth-order linear prediction; (d) twin-R correlative prediction. Peak voltage denotes the location of R waves.

Figure 5. Proposed scheme: (a) original samples (SNR ¼ 10 dB); (b) compressed output via (4, 2) LWT.

of the extracted FECG, (b) illustrates the first-order linear prediction, (c) presents the fourthorder linear prediction, and (d) shows the output of twin-R correlative prediction. We justify that the fourth-order predictor contains more false detections but less average errors comparing to the first-order predictor, while the proposed twin-R correlation predictor shows reduced average errors in contrast to the former two linear predictors [8].

We applied LWT for sequential ECG compression in time interval [400:1000] with additive random white noise (fmSNR ¼ �10 dB). Figure 5 displays the original ECG and its compressed output. Comparing Figure 5(b) to Figure 5(a), we justify that the compressed ECG preserved most details of the original data with mild penalty of energy lost in the amplitude which comes from the quantization error as well as the loss from round-off decomposition in LWT.

Let us employ the compression ratio (CR) and percent root mean-square difference (PRD) [8, 12] to measure the quantitative compression performance on sequential ECE data: the tests as described below randomly selected 24 cases out of 48 from MIT-BIS Arrhythmia Database as testing samples. The proposed scheme recorded PRDs in a range of variable CRs with different quantization coefficients in comparison to those obtained by the Sabah's method.


Table 3. PRD comparison: the proposed scheme versus Sabah's.

of the extracted FECG, (b) illustrates the first-order linear prediction, (c) presents the fourthorder linear prediction, and (d) shows the output of twin-R correlative prediction. We justify that the fourth-order predictor contains more false detections but less average errors comparing to the first-order predictor, while the proposed twin-R correlation predictor shows reduced

Figure 5. Proposed scheme: (a) original samples (SNR ¼ 10 dB); (b) compressed output via (4, 2) LWT.

Figure 4. (a) Originally detected FECG; sequential data output by: (b) single linear prediction; (c) fourth-order linear

prediction; (d) twin-R correlative prediction. Peak voltage denotes the location of R waves.

66 Interpreting Cardiac Electrograms - From Skin to Endocardium

We applied LWT for sequential ECG compression in time interval [400:1000] with additive random white noise (fmSNR ¼ �10 dB). Figure 5 displays the original ECG and its compressed output. Comparing Figure 5(b) to Figure 5(a), we justify that the compressed ECG preserved most details of the original data with mild penalty of energy lost in the amplitude which comes

Let us employ the compression ratio (CR) and percent root mean-square difference (PRD) [8, 12] to measure the quantitative compression performance on sequential ECE data: the tests as described below randomly selected 24 cases out of 48 from MIT-BIS Arrhythmia Database as testing samples. The proposed scheme recorded PRDs in a range of variable CRs with different quantization coefficients in comparison to those obtained by the Sabah's method.

from the quantization error as well as the loss from round-off decomposition in LWT.

average errors in contrast to the former two linear predictors [8].

(a)

(b)

(c)

(d)

We averaged each numerical value of PRD that corresponds to different CR, and enumerated the numbers on both the two schemes in Table 3. From the column comparison, we justify that by gaining the same CR ranging from 2.0 to 19.0, our scheme achieved much smaller PRDs than those of Sabah's, which suggests availability of achieving lower distortion rate by the proposed correlative prediction.

While the CR is under determination for both schemes on compression, let us consider CR as time points and PRD as the sequential output, a GM(1, 1) for the "time-sequence" T1, T<sup>2</sup> is now constructed in order to obtain the predictive value of PRDs. From Eqs. (14) to (16), we are able to justify the positive correlativity between T<sup>1</sup> and T2. From Eqs. (17) to (20), the solution to predictive GM(1, 1) model after LS updates and iterations is formulated as [8]:

$$T\_1: \quad T\_1^{(1)}(k+1) = (2.10 - 328.91)e^{-0.0026784k} + 328.91\tag{28}$$

$$T\_2: \quad T\_2^{(1)}(k+1) = (1.49 - 150.95)e^{-0.0049531k} + 150.95\tag{29}$$

Table 4 illustrates each value of the predicted PRDs obtained by our scheme versus Sabah's in condition of "extrapolated" and "interpolated" CRs. From column comparison, we justify that the GM(1, 1) prediction model performs well for the "extrapolated" CRs and presents closer predicted results in contrast with those of real PRD values in Table 3; most notably, if CRs become large enough, higher order polynomial fittings can be less reliable than predicting the "interpolated" time points while the functional fittings make less sense for extrapolated points, that is an auxiliary reason for using the gray system model on prediction.


Table 4. Predicted PRD (by GM(1,1)) of the proposed scheme versus Sabah's.

The third set of experiments illustrates the simulations of extracting FECG from MECG with additive noise with an adaptive least mean square (LMS) noise canceller to perform this task, which are depicted in Figure 6 as modified from Dr. Igal A. Sebag's example [39] on both maternal and fetal heartbeat detections using sample synthetic ECG data. The six subplots permuted in the top row and in the middle row of Figure 6 show the measuring procedure till the recovery of fetal heartbeat, where the convergence of adaptive noise cancellation takes up to 5–6 s on average. With the assumptions on a sampling rate of 4000 Hz and time duration of 40 s, the maternal heart rate is 89 beats per minute (bpm), and the fmSNR is adjusted as approximately 11.5 dB so as to simulate a test example on the third trimester of pregnancy. The fetal heart rate (FHR) is apparently faster than that of the mother's, normally ranging from 120 to 160 bpm and descending with the progress of gestational weeks. Since the measured FECG via abdominal recordings is often dominated by the maternal heartbeat signal propagating from the chest cavity to maternal abdomen, such path of propagation is constructed as a finite impulse response (FIR) filter with 10 randomized coefficients, with uncorrelated additive random noise which is 0.02 time of the original signal. While the reference signal of MECG is still surrounded with noise, an adaptive LMS filter with 15 coefficients and a step size of 0.00007 can be applied for simplicity of use. Note that the remainder of the error signal after the convergence of the system indicates an estimate of the fetal heartbeat signal associated with the measurement noise.

The bottom row of Figure 6 comprises three subplots, where the left one shows the filtered FECG in contrast to its reference, the middle one indicates peak detection by applying dynamic thresholds to the filtered FECG and using vertical lines to mark each peak on any FECG signal

Figure 6. Automatic fetal heart rate detection on the reconstructed FECG from the original heartbeat signals of both mother and fetus.

crossing the threshold, and the right one depicts the reconstructed FECG data with variations and the automatically calculated FHR equals 135 bpm during the time interval of 36–40 s, which coincides with those normal diagnostic examples in FHR monitoring before delivery.

The third set of experiments illustrates the simulations of extracting FECG from MECG with additive noise with an adaptive least mean square (LMS) noise canceller to perform this task, which are depicted in Figure 6 as modified from Dr. Igal A. Sebag's example [39] on both maternal and fetal heartbeat detections using sample synthetic ECG data. The six subplots permuted in the top row and in the middle row of Figure 6 show the measuring procedure till the recovery of fetal heartbeat, where the convergence of adaptive noise cancellation takes up to 5–6 s on average. With the assumptions on a sampling rate of 4000 Hz and time duration of 40 s, the maternal heart rate is 89 beats per minute (bpm), and the fmSNR is adjusted as approximately 11.5 dB so as to simulate a test example on the third trimester of pregnancy. The fetal heart rate (FHR) is apparently faster than that of the mother's, normally ranging from 120 to 160 bpm and descending with the progress of gestational weeks. Since the measured FECG via abdominal recordings is often dominated by the maternal heartbeat signal propagating from the chest cavity to maternal abdomen, such path of propagation is constructed as a finite impulse response (FIR) filter with 10 randomized coefficients, with uncorrelated additive random noise which is 0.02 time of the original signal. While the reference signal of MECG is still surrounded with noise, an adaptive LMS filter with 15 coefficients and a step size of 0.00007 can be applied for simplicity of use. Note that the remainder of the error signal after the convergence of the system indicates an estimate of the fetal heartbeat signal associated

The bottom row of Figure 6 comprises three subplots, where the left one shows the filtered FECG in contrast to its reference, the middle one indicates peak detection by applying dynamic thresholds to the filtered FECG and using vertical lines to mark each peak on any FECG signal

Figure 6. Automatic fetal heart rate detection on the reconstructed FECG from the original heartbeat signals of both

with the measurement noise.

68 Interpreting Cardiac Electrograms - From Skin to Endocardium

mother and fetus.

The experiments were conducted and retested via the software platform of MATLAB R2011a and higher versions in a Dell laptop with Core i7-4500U 1.80G CPU and 8GB RAM. We plan to include some specific analysis on the single-channel recordings for both healthy and pathological patients during the second and third trimester of pregnancy, and how the theoretical noninvasive FECG extraction algorithms influence the reconstruction accuracy of ECG signals from clinical experiments in later investigations.

Simulations on a real-life monitoring case were included in the two diagrams of Figure 7, where the occurrence of a typical scene on fetal hypoxia was illustrated in types of two parameters such as FHR (ranging from 70 to 150 bpm) and T/QRS (30 samples) in a 10-min

Figure 7. Simulations of sample data on FHR and T/QRS [43] in a 10-min time sequence for cases of: (a) physiological recordings (top); and (b) pathological recordings (bottom).

recording [43]. The top diagram and the bottom diagram depict recordings of a physiological sample and a pathological sample, respectively. Relatively steady FHR in Figure 7(a) indicates fetus in good condition, while the large valley in the waveform of FHR (especially at around the time of minute 06:00) in Figure 7(b) exhibits the abnormal oscillations of fetal heartbeats resulting from intrauterine hypoxia. For the physiological sample, the parameters are comparatively stable in which FHR displays fluctuations around 140 bpm at most of the time and T/ QRS indicates minor numerical vibrations around 0.1; for the pathological sample, FHR exhibits larger amplitude of oscillations ranging from 70 to 140 bpm, while the numerical value of balance point on T/QRS fluctuations is around 0.2. In this trial, it is concluded that the value of fmSNR represents the most crucial factor affecting the quality of filtration, while both the parameter settings on the adaptive filtering system and the locations of electrodes contribute to the signal outputs on abdominal recordings [43].

## 5. Conclusions

A concise study of recent noninvasive FECG detection schemes has been established in this chapter. We have investigated a variety of algorithms for both single-channel and multichannel noninvasive FECG separation from MECG in abdominal recordings. The extended Kalmanbased approach with algorithm variations modeled nonlinearity in single-channel cases, achieved considerably good performance on both synthetic and real-life ECG data. The extended Kalman smoother with ANFIS displays the best overall results on the set of performance metrics among nine noninvasive methods for FECG detection.

We have proposed a scheme of twin-R correlative prediction by applying (4, 2) LWT that effectively exploits correlation characteristics of time-domain ECG for sequential data processing. We have feasibly realized the parameter evaluation of ECG compression by building up a predictive GM(1, 1) model in order to give solutions to PRDs with both extrapolated and interpolated CRs, and achieved lower distortion rates in contrast to those of Sabah's. The correlation predictor with the multivariable gray model displays validity and efficiency for compression, suggesting a prospective technique for ECG data prediction and parameter evaluation. The modified simulation trials on fetal heartbeat detection by reconstructing FECG from maternal abdominal recordings via adaptive noise cancellation, provide an example on automatic FHR estimation on synthetic ECG data [39]; sample trials for either physiological case or pathological case on FHR recordings with modeling of hypoxia on mature fetus were included and reported in adaptive control systems for noninvasive monitoring [43].

As future work, we plan to improve one of the recent noninvasive FECG detection schemes by collaborating high-order dimensional data mining (i.e., inducing robust tensor decompositions to the dynamic filter-based models) to the ICA-based JADE scheme for FECG extraction from multichannel abdominal recordings, and updating the prediction system via statistical machine learning other than ANFIS, where the cocktail party-based solutions suggest a feasible tool for technical improvements [40]. We also plan to design an analytical software platform using the wavelet toolbox, which is oriented for detecting fetal cardiac arrhythmias with more practical trials on real data toward the multilead system for abdominal ECG recordings.

## Author details

#### Xin Gao

recording [43]. The top diagram and the bottom diagram depict recordings of a physiological sample and a pathological sample, respectively. Relatively steady FHR in Figure 7(a) indicates fetus in good condition, while the large valley in the waveform of FHR (especially at around the time of minute 06:00) in Figure 7(b) exhibits the abnormal oscillations of fetal heartbeats resulting from intrauterine hypoxia. For the physiological sample, the parameters are comparatively stable in which FHR displays fluctuations around 140 bpm at most of the time and T/ QRS indicates minor numerical vibrations around 0.1; for the pathological sample, FHR exhibits larger amplitude of oscillations ranging from 70 to 140 bpm, while the numerical value of balance point on T/QRS fluctuations is around 0.2. In this trial, it is concluded that the value of fmSNR represents the most crucial factor affecting the quality of filtration, while both the parameter settings on the adaptive filtering system and the locations of electrodes contribute to

A concise study of recent noninvasive FECG detection schemes has been established in this chapter. We have investigated a variety of algorithms for both single-channel and multichannel noninvasive FECG separation from MECG in abdominal recordings. The extended Kalmanbased approach with algorithm variations modeled nonlinearity in single-channel cases, achieved considerably good performance on both synthetic and real-life ECG data. The extended Kalman smoother with ANFIS displays the best overall results on the set of perfor-

We have proposed a scheme of twin-R correlative prediction by applying (4, 2) LWT that effectively exploits correlation characteristics of time-domain ECG for sequential data processing. We have feasibly realized the parameter evaluation of ECG compression by building up a predictive GM(1, 1) model in order to give solutions to PRDs with both extrapolated and interpolated CRs, and achieved lower distortion rates in contrast to those of Sabah's. The correlation predictor with the multivariable gray model displays validity and efficiency for compression, suggesting a prospective technique for ECG data prediction and parameter evaluation. The modified simulation trials on fetal heartbeat detection by reconstructing FECG from maternal abdominal recordings via adaptive noise cancellation, provide an example on automatic FHR estimation on synthetic ECG data [39]; sample trials for either physiological case or pathological case on FHR recordings with modeling of hypoxia on mature fetus were included and reported in adaptive

As future work, we plan to improve one of the recent noninvasive FECG detection schemes by collaborating high-order dimensional data mining (i.e., inducing robust tensor decompositions to the dynamic filter-based models) to the ICA-based JADE scheme for FECG extraction from multichannel abdominal recordings, and updating the prediction system via statistical machine learning other than ANFIS, where the cocktail party-based solutions suggest a feasible tool for technical improvements [40]. We also plan to design an analytical software platform using the wavelet toolbox, which is oriented for detecting fetal cardiac arrhythmias with more practical trials on real data toward the multilead system for abdominal ECG recordings.

mance metrics among nine noninvasive methods for FECG detection.

the signal outputs on abdominal recordings [43].

70 Interpreting Cardiac Electrograms - From Skin to Endocardium

control systems for noninvasive monitoring [43].

5. Conclusions

Address all correspondence to: xgao1985@email.arizona.edu

Department of Electrical and Computer Engineering, the University of Arizona, Tucson, USA

## References


[25] Rosén KG, Samuelsson A. Device for reducing signal noise in a fetal ECG signal. U.S. Patent 6658284, issued December 2, 2003

[11] Immanuel JJR, Prabhu V, Christopheraj VJ, Sugumar D, Vanathi PT. Separation of maternal and fetal ECG signals from the mixed source signal using FASTICA. Procedia Engi-

[12] Jafari MG, Chambers JA. Fetal electrocardiogram extraction by sequential source separation in the wavelet domain. IEEE Transactions on Biomedical Engineering. 2005;52

[13] Jalaleddine SMS, Hutehens CG, Strattan RD, Coberly WA. ECG data compression techniquesA unified approach. IEEE Transactions on Biomedical Engineering. 1990;37

[14] Kropfl M, Modre-Osprian R, Schreier G, Hayn D. A robust algorithm for fetal QRS detection using non-invasive maternal abdominal ECGs. Computing in Cardiology.

[15] Kumar P, Sharma SK, Prasad S. Detection of FECG from multivariate abdominal recordings using wavelets and neuro-fuzzy systems. International Journal of Engineering and

[16] Lathauwer L, Moor B, Vanderwalle J. Fetal electrocardiogram extraction by blind source subspace separation. IEEE Transactions on Biomedical Engineering. 2000;47(5):567–572

[17] Guerrero-Martinez JF, Martinez-Sober M, Bataller-Mompean M, Magdalena-Benedito JR. New algorithm for fetal QRS detection in surface abdominal records. Computers in

[18] Melillo P, Santoro D, Vadursi M. Detection and compensation of inter-channel time offsets in indirect fetal ECG sensing. IEEE Sensors Journal. 2014;14(7):2327–2334

[19] Panigrahy D, Sahu PK. Extraction of fetal electrocardiogram (ECG) by extended state Kalman filtering and adaptive neuro-fuzzy inference system (ANFIS) based on single

[20] Peters CHL, Van Laar JOEH, Vullings R, Oei SG, Wijn PFF. Beat-to-beat heart rate detection in multi-lead abdominal fetal ECG recordings. Medical Engineering & Physics.

[21] Poian GD, Bernardini R, Rinaldo R. Separation and analysis of fetal-ECG signals from compressed sensed abdominal ECG recordings. IEEE Transactions on Biomedical Engi-

[22] Reza S, Shamsollahi MB, Jutten C, Clifford GD. A nonlinear Bayesian filtering framework for ECG denoising. IEEE Transactions on Biomedical Engineering. 2007;54(12):2172–2185

[23] Rooijakkers MJ, Rabotti C, de Lau H, Oei SG, Bergmans JWM, Mischi M. Feasibility study of a new method for low-complexity fetal movement detection from abdominal ECG recordings. IEEE Journal of Biomedical and Health Informatics. 2016;20(5):1361–1368 [24] Rosén KG, Amer-Wåhlin I, Luzietti R, Norén H. Fetal ECG waveform analysis. Best

Practice & Research Clinical Obstetrics & Gynaecology. 2004;18(3):485–514

channel abdominal recording. Sadhana. 2015;40(Part 4):1091–1104

neering. 2012;30:356–363

72 Interpreting Cardiac Electrograms - From Skin to Endocardium

(3):390–400

(4):329–343

2013;40:313–316

Cardiology. 2006;33:441–444

2012;34(3):333–338

neering. 2016;63(6):1269–1279

Advanced Technology Studies. 2013;2(1):45–51


**Understanding Electrograms from Cardiac Devices**

[40] https://cran.r-project.org/web/packages/JADE/vignettes/JADE-BSSasymp.pdf

8220/17/5/1154

74 Interpreting Cardiac Electrograms - From Skin to Endocardium

[41] https://www.cablesandsensors.com/pages/12-lead-ecg-placement-guide-with-llustrations [42] Lynn WD, Escalona OJ, McEneaney DJ. Arm and wrist surface potential mapping for wearable ECG rhythm recording devices: A pilot clinical study. Sensors & their Applications, Journal of Physics: Conference Series. 2013;450:012026. DOI: 10.1088/1742-6596/ 450/1/012026; http://iopscience.iop.org/article/10.1088/1742-6596/450/1/012026/pdf [43] Martinek R, Kahankova R, Nazeran H, Konecny J, Jezewski J, Janku P, Bilik P, Zidek J, Nedoma J, Fajkus M. Non-invasive fetal monitoring: A maternal surface ECG electrode placement-based novel approach for optimization of adaptive filter control parameters using the LMS and RLS algorithms. Sensors. 2017;17(5):1154. http://www.mdpi.com/1424-

**Provisional chapter**
