الفهرس 
Chapter 1 IntroductionOnly 14 pages are availabe for public view
 |  | from | 98 |  |  |
| | | from | 98 | | |
Abstract
Cardiovascular diseases are currently the biggest single cause of death in developed countries, so the development of better diagnostic methodologies could improve the health of many people. Arrhythmias are related to the sudden cardiac death, one of the challenges for the modern cardiology.
On the other hand, the classification of heartbeats on the electrocardiogram (ECG) is more important analysis than the study of arrhythmias. The automation of heartbeat classification could improve the diagnostic quality of arrhythmias, especially in long-term recordings. The objective of this thesis is the study of the methodologies for the classification of heartbeats on the ECG.
We developed and validated a simple heartbeat classifier based on features extracted with Legendre Moments and classified with using Multiclass SVM (ECOC).
First, we created and approved a simple heartbeat classifier dependent on features chose with the focus on an improved generalization capability. We thought about features from the RR intervals (remove between two back to back pulses) arrangements, just as features processed from the ECG tests and from scales of the wavelet transform, at both available leads. The classification performance and generalization have been studied by using public available databases: the MIT-BIH Arrhythmia, available in Physionet. The main idea of the proposed frame work is to classify the ECG beats to main four types. These types are normal beat (normal), Left Bundle Branch Block beats (LBBB), Right Bundle Branch Block beats (RBBB), Atrial Premature Contraction (APC).
The best performing strategy consisted in performing with using Legendre Moments as feature extractor and Multiclass SVM (ECOC) as a classifier. . Simulation results approved that the proposed approach gives 97.7% accuracy levels compared to 95.7447%, 95.88%, 95.03% , 93.40%, 96.02%, 95.95%, 96.24% achieved with Discrete wavelet (DWT), Haar wavelet and principle
iii
component analysis (PCA) as feature extractors and ANN, Simple Logic Random Forest, SVM and J48 as classifiers.
In the second strategy, we have used classification with Convolutional Neural Network, The importance of ECG classification is very high now due to many current medical applications where this problem can be stated. Currently, there are many machine learning (ML) solutions which can be used for analyzing and classifying ECG data. However, the drawbacks of these ML results is the usage of heuristic hand-crafted or engineered features with shallow feature learning architectures. The problem relies in the possibility not to find most appropriate features which will give high classification accuracy in this ECG problem. One of the proposed solutions is to use deep learning architectures where first layers of convolutional neurons behave as feature extractors and in the end some fully-connected (FCN) layers are used for making final decision about ECG classes. CNNs are performed on fundus images to classify the normal and abnormal cases. The accuracy of this approach is (100 %).
The accurate detection and classification are our target, and accurate classification for the ECG beats saves the life for many people and it’s good helper for heart specialists. Saving human life is the main target of all this efforts and work.