Methodology for Classification and Analysis of Neonate and Adult ECG

Abstract

The electrocardiogram (ECG) is widely used for diagnosis of cardiopathies and others diseases. The analysis of the ECG may be carried out by computational systems. However, algorithms developed for analysis of the ECG of adult patients causes false alarms when applied to neonatal arrhythmias monitoring, due to the differences between the characteristics of the adult ECG in comparison to the normal ECG of neonatal patients. The main differences between the adult and neonate ECG are due to higher heart rate and right ventricle dominance observed in the newborn patients. These differences should be considered during the development of a computational system for ECG automated interpretation. The aim of this work is to present an efficient methodology to analyze and to classify the ECG of adults and neonates. Thus, ECG can be monitored for arrhythmia detection. The proposed methodology is divided into two steps (1) heartbeats are classified by an artificial neural network (ANN) among normal, premature ventricular contraction and bundle-branch block. The ANN architecture has the input layer composed by 225 samples of the QRS wavelet transform and the output layer has three neurons (one for each heartbeat class); (2) information about ECG intervals duration and heart rate are applied in an expert system that considers patient age to diagnose arrhythmias related to the atrial and ventricular rhythm (arrhythmic episodes). The ECG information is compared to the expected values for the patient age. So, the ECG episodes can be classified as normal, bradicardia, tachycardia, AV block, ventricular tachycardia, and QT interval abnormalities (like Short QT Syndrome or Long QT Syndrome). The MITBIH arrhythmia database and ECG files recorded from newborn patients were used for evaluation of the implemented methodology. The results reached a sensitivity of 92.5% and specificity of 96.7% showing the efficiency of the developed methodology for ECG analysis and classification of neonates and adults patients.