Automated ECG beat classification using DWT and Hilbert transform-based PCA-SVM classifier

The analysis of electrocardiogram (ECG) signals provides valuable information for automatic recognition of arrhythmia conditions. The objective of this work is to classify five types of arrhythmia beat using wavelet and Hilbert transform-based feature extraction techniques. In pre-processing, wavelet transform is used to remove noise interference in recorded signal and the Hilbert transform method is applied to identify the precise R-peaks. A combination of wavelet, temporal and morphological or heartbeat interval features has been extracted from the processed signal for classification. The principal component analysis (PCA) is used to select the informative features from the extracted features and fed as input to the support vector machine (SVM) classifier to classify arrhythmia beats automatically. We obtained better performance results in the PCA-SVM-based classifier with an average accuracy, sensitivity and specificity of 98.50%, 95.68% and 99.18%, respectively in cubic-SVM classifier for classifying five types of ECG beats at fold eight in ten-fold cross validation technique. The effectiveness of our method is found to be better compared to published results; therefore, the proposed method may be used efficiently in the ECG analysis.