A Wavelet-Based ECG Delineation and Automated
Diagnosis of Myocardial Infarction in PTB Database
Rachid HADDADI *, Elhassane ABDELMOUNIM*, Mustapha EL HANINE*,
Abdelaziz BELAGUID**
haddadirachid@hotmail.com, hassan.abdelmounim@hotmail.fr, melhanine@yahoo.fr
* Laboratory of Systems Analysis and Information Processing, Faculty of Sciences and Technics, Hassan
first University, Settat, Morocco.
** Laboratory of physiology, Mohammed-V University Rabat, Morocco
Abstract. In this work, we present an ECG delineation and the automated diagnosis of
coronary artery disease in the electrocardiogram (ECG). In preprocessing stage, the
baseline wander (BLW) and 60 Hz power line interference (PLI) were removed using
discrete wavelet transform (DWT). The QRS detection is carried out using Daubechies
(Db4) DWT. Feature extraction and classification is done using a convolutional neural
network (CNN) containing three convolutional layers, three max-pooling layers, and three
fully connected layers. The standard 12 lead ECG signals of 50 healthy subjects and 50
myocardial infarction subjects (MI) of one minute are obtained from the Physikalisch-
Technische Bundesanstalt (PTB) database. We achieved an accuracy of 94.83%.
sensitivity of 94.75%, and specificity of 94.93% on PTB database.
Keywords: Electrocardiogram, DWT, QRS complex, Convolutional neural network.
1 Introduction
Cardiovascular disease is the leading cause of death in the world. According to the World
Health Organization (WHO), more than 17.7 million people die each year due to cardiovascular
disease, representing 31% of total global mortality [1]. This number is in permanent growth.
The main risk factors are obesity, tobacco, high blood pressure, diabetes and genetic factors.
Ischemic heart disease happens when coronary arteries get narrower and reduce the blood flow
to the heart. This is also called coronary heart disease (CHD). This can ultimately lead to the
heart attack [2]. The electrocardiogram (ECG) is the most used cardiological examination at
hospitals and cardiology offices to diagnose heart disease. It is easy to set up, painless and
inexpensive.
As illustrated in Figure 1, an ECG is a series of waves and deflections recording the heart’s
electrical activity from a certain view. Each view called a lead, monitor voltage changes between
electrodes placed in different positions on the body. Leads I, II, and III are bipolar leads. Leads
aVR, aVL, aVF, V1 through V6 are unipolar leads. An ECG tracing looks different in each lead
because the recorded angle of electrical activity changes with each lead [3]. The ECG signal is
characterized by a recurrent wave sequence of P, QRS, and T-wave associated with each beat.
Each wave of the ECG represents a phase of heart functioning. Each deformation detected on
the duration and the forms of these waves can be considered as an indicator of cardiac
abnormalities.
ICCWCS 2019, April 24-25, Kenitra, Morocco
Copyright © 2019 EAI
DOI 10.4108/eai.24-4-2019.2284216