Automatic detection of characteristic segments of a recorded ECG signal with noise handling methods

Abstract

<jats:title>Abstract</jats:title> <jats:p>Depolarization and repolarization of the myocard results in specific wave shapes that are recognizable in an ECG signal and are characterized by their length, rise time and amplitude. Changes in these characteristics usually indicate anomalies in the function of the heart. Automatic detection of characteristic segments of the ECG signal is of crucial importance for fast and reliable recognition of artefacts that are further analyzed as a means for setting a diagnosis. In this paper a contribution has been made towards efficient automatic segment detection of real ECG signals recorded in arbitrary conditions in the presence of noise from various sources. The method of accumulated differential and a technique for tracing the wave shapes of the ECG signal using their local extrema have been implemented in the developed algorithm. The implementation of these methods yields an advanced way of handling the noise present in the ECG signal resulting with high precision segment detection and low computational power requirement. The high accuracy of the algorithm has been verified using real signals recorded on 2 lead ambulatory electrocardiograph. Another benefit of the developed algorithm is its speed – it is able to process a 5 minute long recorded ECG signal in 9 seconds.</jats:p>