Prediction and detection of premature ventricular contractions based on the electrocardiogram signal processing and neural network

rahman amini¹ mehdi moghzi²

1) 1. Department of Engineering of Radiation Medicine, Islamic Azad University Isfahan Khorasgan Branch
2) 1. Department of Engineering of Radiation Medicine, Islamic Azad University Isfahan Khorasgan Branch

Publication : 2nd. International Congress on science & Engineering - paris(parisconf.com)
Abstract :
Cardiovascular disease is one of the leading causes of death in today s industrialized world, and ventricular fibrillation is the most commonly encountered, accounting for about 1 percent of the general population in the whole world. The incidence of this disease increases with age. Diagnosis of cardiac arrhythmias is important and important. Because long-term arrhythmias create more risks for the individual and can even lead to the death of the patient. Therefore, the diagnosis of arrhythmias should be of high accuracy so as to be able to track arrhythmia levels. For some time now, ways to diagnose and classify these arrhythmias have been a major problem for cardiologists. Early ventricular contractions (PVC) are one of the most common heart rate abnormalities. Early ventricular contractions are irregular activity that occurs due to severe ventricular contractions, and thus abnormal patterns occur in the QRS complex. These repeated contractions during the physical activity increase the risk of death and can lead to more serious cardiac arrhythmia, such as atrial fibrillation. Early ventricular contractions are easily detected in an ECG with a distinct shape. For this purpose, a Monitoring Holter can be used to investigate the morphology of the signals. In recent decades, many studies have been done to identify patients with premature ventricular contractions using electrocardiogram signals. In this study, time intervals features and neural network, SVM and KNN classifier for prediction of premature ventricular contractions used. The results showed that the single-layer neural network with 30 neurons in the middle layer and KNN classifier has a higher performance for diagnosis of healthy. On average, the KNN classification has performance in detecting the risk levels of the PVC disease.
Keywords : ECG signal premature ventricular contractions Feature Extraction Classification