Detection and extraction of the ECG signal parameters

Abstract

This work investigates a set of efficient techniques to extract important features from the ECG data applicable in automatic cardiac arrhythmia classification. The selected parameters are divided into two main categories namely morphological and statistical features. Extraction of morphological features was achieved using signal processing techniques and detection of statistical features was performed by employing mathematical methods. Each specific method was applied to a pre-selected data segment of the MIT-BIH database. The classification of different heart beats was performed based upon the extracted features. The morphological features were found as the most efficient for further ECG signal analysis. However, because of ECG signal variability in different patients, the mathematical approach is preferred for a precise and robust feature extraction. As a result of the extracted features, an efficient computer based ECG signal classifier could be developed for detection of a vast range of cardiac arrhythmias.