ELECTROCARDIOGRAM SIGNAL ANALYSIS: DETECTION AND LOCALIZATION OF ECG COMPONENTS

This paper proposes a method for the detection and localization of characteristic elements of electrocardiographic signals based on a nonlinear filtering scheme (NLFS). The developed approach is intended for the automatic detection of the main components of a single-channel ECG and consists of two stages. At the first stage, a mathematical model was developed to identify characteristic points of a synthetic ECG signal without noise, including the QRS complex peak as well as the onset and offset points of the P and T waves. At the second stage, the proposed model was adapted for processing real ECG signals containing noise and artifacts. The performance of the proposed method was evaluated using the standard QT-MIT database. For QRS complex peak detection, the method achieved a sensitivity (Se) of 98.88% and a positive predictive value (P+) of 98.43%. The sensitivity for detecting the onset of the P wave, the offset of the P wave, and the offset of the T wave was 75.16%, 71%, and 90.7%, respectively. In addition, the mean values and standard deviations of the differences between the automatic detection results and expert annotations were calculated. The obtained results confirm the applicability of the proposed approach for automated analysis of electrocardiographic signals.