ECG Signals Classification using Statistical and Time-Frequency Features

Cardiovascular diseases are one of the most frequent and dangerous problems in modern society nowadays. Therefore, it is very difficult to take immediate measures without real time electrocardiogram information. Unfortunately ECG signals, during their acquisition process, are affected by various types of noise and artifacts due to the movement, or breathing of the patient, electrode contact, power-line interferences, etc. The aim of this study was to develop an algorithm to detect and classify four types of electrocardiograms (ECG): without noise, or containing one of the following three types of noise: baseline wonder, muscular noise or electrode motion artifact. The classification was made using descriptive statistics. The Stationary Wavelet Transform (SWT) was applied in order to extract features from input signals. The main reasons for using this transform are the properties of good representation of non-stationary signals such as ECG signals and the possibility of dividing the signal into different bands of frequency. The proposed method was tested using real ECG signals affected by noise from the MIT-BIH arrhythmia database. The goal was to analyze the percentage of the well classified signals. The proposed algorithm showed good results, assuring a good classification with more than 90% well classified signals for each type of ECGs.