Automatic Arrhythmia Detection

Automatic Arrhythmia Detection

Carlos M. Travieso (University of Las Palmas de Gran Canaria, Spain), Jesús B. Alonso (University of Las Palmas de Gran Canaria, Spain), Miguel A. Ferrer (University of Las Palmas de Gran Canaria, Spain) and Jorge Corsino (University of Las Palmas de Gran Canaria, Spain)
DOI: 10.4018/978-1-61520-893-7.ch013

Abstract

In the present chapter, the authors have developed a tool for the automatic arrhythmias detection, based on time-frequency features and using a Support Vector Machines (SVM) as classifier. Arrhythmia Database Massachusetts Institute of Technology (MIT) has been used in the work in order to detect eight different states, seven are pathologies and one is normal. The unions of different blocks and its optimization have found success rates of 99.82% for RR’ interval detection from electrocardiogram (PQRST waves), and 99.23% for pathologic detection. In particular, the authors have used wavelet transform in order to characterize the wave of electrocardiogram (ECG), based on Biorthogonal family, achieving the most discriminative coefficients. A discussion on arrhythmia ECG classification methods is also presented in this paper.
Chapter Preview
Top

Background

In the 60’s the ECG analysis had a wide number of studies about its process in different ways and for different applications, for example, some of them apply mathematical or statistical concepts.

Most cardiovascular diseases can be diagnosed by the analysis of the significant time intervals in the ECG signal. The ECG features extracted can determine whether a patient presents any kind of cardiac pathology or not. However, in order to extract these values a previous signal processing is required to delete noise and to make the analysis lighter. The common noise sources are: baseline wander, power line interference and muscle noise. Baseline wander, or stranger low-frequency high-bandwidth components, can be caused by: perspiration (effects electrode impedance), respiration and body movements. These noises can cause problems to analysis, especially when examining the low-frequency ST-T segment.

On the frequency domain, the main information is placed under 35 Hz. Power line normally affect over 50 and 60 Hz and the muscle noisy can be approximated as white noise because his frequency spectrum is practically plane.

The inherent morphology of the electrocardiographic (ECG) signal is a key factor to determine if the reliable feature detection can be made. As a non-stationary signal, ECG has to be pre-processed using frequency adaptive tools. Several techniques have been used to perform the detection of these features over the last decades, such as the Short Time Fourier Transform (STFT) or the Wavelet transform.

In the arrhythmias detection there are two main kinds for the feature extraction. On the one hand, features extracted directly from the ECG signal, time, sizes, areas, delays, etc. And on the other hand, features extracted from domain transforms. Both methods will be discussed in this present work (Mallat, 1998; Kunzmann et al. 2002; Alonso et al., 1999; Moraes et al., 2002; Bahoura et al., 1997; Hosseini & Reynolds, 2001; Zimmerman & Povinelli, 2004).

Complete Chapter List

Search this Book:
Reset