Automated detection of cardiac health condition using linear techniques

Abstract

Coronary Artery Disease (CAD) is the basic source for various cardiac diseases such as Myocardial Infraction (MI) or Heart Attack (HA), Heart Failure (HF) and Ischemic Heart Disease (IHD). Detection of CAD in the early stages and treatment for the same will help in preventing from continuing further. However earlier and precise recognition of CAD from Electrocardiogram (ECG) signals using manual recognition is not an easy job to perform. Thus computer aided techniques are essential for the automatic classification of CAD condition. This work proposes an automated classification of normal and CAD conditions of ECG using linear techniques. In this method normal and CAD ECG beats are de-noised using Discrete Wavelet Transform (DWT) technique. Feature extraction of ECG signals is achieved by Discrete Cosine Transform (DCT). Dimensionality reduction of data set is done by Principle Component Analysis (PCA). The resulting PCA features are partitioned using Cross Validation Partition for the classification purpose. Normal and CAD ECG signals are classified by k-Nearest Neighbor (k-NN) classifier and obtained results are compared with Support Vector Machine (SVM) classifier.