Synthesis of neural network approaches used for ECG classification

Cardiovascular diseases represent the most frequent cause of death in Europe. Consequently, their diagnoses appear a vital task. In cardiology unit, ECG signal still remains the dominant used tools for arrhythmia and heart diseases analysis. In fact, the ECG is a non-invasive tool to explore the functional heart sate. It is an electrical signal varying according to the electrical heart state. From the ECG signal, some significant parameters can be extracted. Generally, the durations and the shapes of the different waves are taken as bio-indicators of certain cardiac anomalies. In fact, manual detection and classification of ECG waves is a difficult and annoying task especially for the analysis of the long recordings as in Holters and ambulatory cases. Besides, a detailed analysis of 12-leads ECG used to identify the presence or the absence of the cardiac malfunction is also irritating. However, and due to large number of patients in intensive care units, the need for continuous heart activity monitoring is necessary requiring therefore the automatic analysis of the ECG signal.

Autorentext

Radhwane Benali received the engineering & the MSc degrees de in Biomedical electronics from the UABB University of Tlemcen, Algeria in 2005 & 2008, respectively. Currently, he is preparing a PHD at Tlemcen University & he is a member of Biomedical Engineering Laboratory.His area of research interests includes biomedical signal processing.