ECG signal processing for abnormalities detection using multi-resolution wavelet transform and Artificial Neural Network classifier

This paper deals with ECG signal analysis based on Artificial Neural Network and combined based (discrete wavelet transform and morphology) features. We proposed a technique to truthfully classify ECG signal data into two classes (abnormal and normal class) using various neural classifier. MIT-BIH arrhythmia database utilized and selected 45 files of one minute recording (25 files of normal class and 20 files of abnormal class) out of 48 files based on types of beat present in it. The total 64 features are separated in to two classes that is DWT (48) based features and morphological (16) feature of ECG signal which is set as an input to the classifier. Three neural network classifiers: Back Propagation Network (BPN), Feed Forward Network (FFN) and Multilayered Perceptron (MLP) are employed to classify the ECG signal. The classifier performance is measured in terms of Sensitivity (Se), Positive Predictivity (PP) and Specificity (SP). The system performance is achieved with 100% accuracy using MLP.