Deep Learning-Based Approach for Atrial
Fibrillation Detection
Lazhar Khriji
1,2(&) , Marwa Fradi
2 , Mohsen Machhout
2 ,
and Abdulnasir Hossen
1
1 College of Engineering, Sultan Qaboos University, Muscat, Oman
{lazhar,abhossen}@squ.edu.om
2 Faculty of Sciences of Monastir, Monastir University, Monastir, Tunisia
marwafradi32@gmail.com, machhout@yahoo.fr
Abstract. Atrial Fibrillation (AF) is a health-threatening condition, which is a
violation of the heart rhythm that can lead to heart-related complications.
Remarkable interest has been given to ECG signals analysis for AF detection in
an early stage. In this context, we propose an artificial neural network ANN
application to classify ECG signals into three classes, the first presents Normal
Sinus Rhythm NSR, the second depicts abnormal signal with Atrial Fibrillation
(AF) and the third shows noisy ECG signals. Accordingly, we achieve 93.1%
accuracy classification results, 95.1% of sensitivity, 90.5% of specificity and
98%. Furthermore, we yield a value of zero error and a low value of cross
entropy, which prove the robustness of the proposed ANN model architecture.
Thus, we outperform the state of the art by achieving high accuracy classification without pre-processing step and without high level of feature extraction,
and then we enable clinicians to determine automatically the class of each
patient ECG signal.
Keywords: ECG-classification Á AF detection Á Confusion matrix Á ROC Á
ANN Á Histogram error
1 Introduction
ECG signals classification is a crucial step to determine given the importance to assign
to each patient its ECG class. Indeed, heart diseases have known a big spread in the last
recent years. Such as arrhythmia cardiac problems like Atrial Fibrillation (AFIB). The
prevalence of arterial fibrillation (AF) is increasing during the last few years and
presenting the most common health problem in many countries [1]. AF presents a very
critical health issue, which affects the quality of life of persons and leading to many
risks such as cardiac stroke. An analysis of AF is based on a clinical evaluation and
requires electrocardiogram (ECG) documentation during the arrhythmia. During the
last few years, deep learning (DL) revolutionized the medical area as the deep neural
networks presented the state of the art results in many applications such as computer
vision, image processing, robotics, medical imaging, etc. The high performance
obtained by the deep neural network is based on the use of powerful graphic processing
units (GPUs) which allowed these implementations to outperform the classic ones.
© The Author(s) 2020
M. Jmaiel et al. (Eds.): ICOST 2020, LNCS 12157, pp. 100–113, 2020.
https://doi.org/10.1007/978-3-030-51517-1_9
Fibrillation Detection
Lazhar Khriji
1,2(&) , Marwa Fradi
2 , Mohsen Machhout
2 ,
and Abdulnasir Hossen
1
1 College of Engineering, Sultan Qaboos University, Muscat, Oman
{lazhar,abhossen}@squ.edu.om
2 Faculty of Sciences of Monastir, Monastir University, Monastir, Tunisia
marwafradi32@gmail.com, machhout@yahoo.fr
Abstract. Atrial Fibrillation (AF) is a health-threatening condition, which is a
violation of the heart rhythm that can lead to heart-related complications.
Remarkable interest has been given to ECG signals analysis for AF detection in
an early stage. In this context, we propose an artificial neural network ANN
application to classify ECG signals into three classes, the first presents Normal
Sinus Rhythm NSR, the second depicts abnormal signal with Atrial Fibrillation
(AF) and the third shows noisy ECG signals. Accordingly, we achieve 93.1%
accuracy classification results, 95.1% of sensitivity, 90.5% of specificity and
98%. Furthermore, we yield a value of zero error and a low value of cross
entropy, which prove the robustness of the proposed ANN model architecture.
Thus, we outperform the state of the art by achieving high accuracy classification without pre-processing step and without high level of feature extraction,
and then we enable clinicians to determine automatically the class of each
patient ECG signal.
Keywords: ECG-classification Á AF detection Á Confusion matrix Á ROC Á
ANN Á Histogram error
1 Introduction
ECG signals classification is a crucial step to determine given the importance to assign
to each patient its ECG class. Indeed, heart diseases have known a big spread in the last
recent years. Such as arrhythmia cardiac problems like Atrial Fibrillation (AFIB). The
prevalence of arterial fibrillation (AF) is increasing during the last few years and
presenting the most common health problem in many countries [1]. AF presents a very
critical health issue, which affects the quality of life of persons and leading to many
risks such as cardiac stroke. An analysis of AF is based on a clinical evaluation and
requires electrocardiogram (ECG) documentation during the arrhythmia. During the
last few years, deep learning (DL) revolutionized the medical area as the deep neural
networks presented the state of the art results in many applications such as computer
vision, image processing, robotics, medical imaging, etc. The high performance
obtained by the deep neural network is based on the use of powerful graphic processing
units (GPUs) which allowed these implementations to outperform the classic ones.
© The Author(s) 2020
M. Jmaiel et al. (Eds.): ICOST 2020, LNCS 12157, pp. 100–113, 2020.
https://doi.org/10.1007/978-3-030-51517-1_9