Multiresolution wavelet transform based feature extraction and ECG classification to detect cardiac abnormalities

Analysis of electrocardiogram (ECG) signal provides valuable information about the heart conditions of the patient to the clinicians. The wavelet transform is an effective tool for extracting discriminative features in ECG signal classification for automatic diagnosis of cardiac arrhythmia. In this paper we proposed an improved algorithm to detect QRS complex features based on the multiresolution wavelet transform to classify four types of ECG beats. Extracted features are used for classifying cardiac abnormalities: normal (N), left bundle branch block (LBBB), right bundle branch block (RBBB), Paced beats (P) using neural network (NN) and support vector machines (SVM) classifier. The performance of the method is evaluated in terms of sensitivity, specificity, accuracy for 48 recorded ECG signals obtained from the MIT-BIH arrhythmia database. The proposed process achieved high detection performance with less error rate of 0.42% in detecting QRS complex. The classifier confirmed its superiority with an average accuracy of 96.67% and 98.39% in NN and SVM respectively. The classification accuracy of SVM approach proves superior for the proposed method to that of the NN classifier with extracted parameter in detecting ECG arrhythmia beats.