Electrocardiogram Signal Analysis Using Artificial Neural Network

Abstract

The electrocardiogram (ECG) is the electrical manifestation of the contractile activity of the heart and can be
recorded fairly easily with surface electrodes on the limbs and chest. It is the most commonly known, recognised
and used biomedical signal. ECG wave shape is altered by cardiovascular diseases and abnormalities such as
myocardial ischemia and infarction, ventricular hypertrophy, and conduction problems. For the ECG analysis, the
method adopted involved extracting features that represent the ECG signals. Eight sets of ECG signals were used.
This was achieved by extracting the QRS complexes within the ECG data first and finally using feature extraction
scheme, to extract key features: Spectral Entropy, Pointcare plot geometry and Largest Lyapunov Exponent (LLE)
that were used to train an Artificial Neural Network (ANN) model. The ANN model thereafter classified the ECG
signals into eight key classes. The analysis showed a very good match from the extracted features after training
the ANN model. The chosen features gave a 100% match when tested against known ECG data samples.
Performance analysis was performed using a confusion matrix to describe how well the classification model
performs in classifying the ECG data. The study was able to achieve the set objective in classifying the cardiac
disorders correctly into their respective classes; showing 90.6% and 97.7% accuracies for two-thirds and 90% of
data used, respectively.