Classification and Feature Extraction

Classification and Feature Extraction

DOI: 10.4018/978-1-5225-5580-3.ch009


In this chapter, the proposed optimization algorithm, kinetic gas molecule optimization (KGMO), that is based on swarm behaviour of gas molecules is applied to train a feedforward neural network for classification of ECG signals. Five types of ECG signals are used in this work including normal, supraventricular, brunch bundle block, anterior myocardial infarction (Anterior MI), and interior myocardial infarction (Interior MI). The classification performance of the proposed KGMO neural network (KGMONN) was evaluated on the Physiobank database and compared against conventional algorithms.
Chapter Preview

9.2 Features Of Disorders

Consultation with heart specialists determined that 4 types of heart disorders, namely, bundle branch block, supraventricular tachycardia, anterior myocardial infarction (anterior MI) and inferior myocardial infarction (inferior MI) are common and would be detected in this work. Sample ECG signals for the 4 classes and normal ECG signals are given in Figure 1. Specialists detect these disorders by observing the PQRST waveform. For example, bundle branch block causes a widened and possibly jagged QRS waveform, while supraventricular tachycardia typically exhibits a narrow QRS complex on the ECG. The ST segment, which is normally iso-electric (flat and in line with the PQ segment) may be elevated or depressed due to myocardial ischemia or myocardial infarction. Table 1 shows the ECG classes and representation of the desired neural network outputs for each class.

Figure 1.

Typical waveform of ECG signals

Table 1.
ECG classes and representation of desired neural network outputs
DisorderClassNeural Network Output
Normal sinus0000
Bundle branch block1001
Supraventricular tachycardia2010
Anterior myocardial infarction3011
Inferior myocardial infarction4100

Complete Chapter List

Search this Book: