A Neural Network Approach for Cardiac Arrhythmia Classification

Rapid or slow heartbeats cause irregular rhythms resulting in Cardiac Arrhythmia, which is assessed by electrocardiogram (ECG). There are various types of arrhythmia and its detection is relevant to heart disease diagnosis. Automatic arrhythmia ECG assessment is a well-researched area. This paper investigates ECG classification using soft computing techniques to classify arrhythmia type through the use of RR interval. Discrete Cosine Transform (DCT) is used to extract features from the time series ECG data using the distance between RR waves. The extracted beat RR interval is used as a feature extracted in the frequency domain and classified using Multi-Layer Perceptron Neural Network (MLP –NN), and proposed Feed Forward Neural Network (FNN) experiments were conducted through the MIT-BIH arrhythmia database.