Authors : Benali Medjahed Oussama, Bachir M`Hamed Saadi and Hadj Slimane Zine-Eddine
Abstract: Electrocardiogram (ECG) is today one of the essential pillars of the diagnosis of heart problems. The analysis of this signal and the identification of its parameters is an important step for diagnosis. In this study, we present a new algorithm for ECG signal classification. Respiratory signal simultaneously recorded with the ECG signal will be used to classify each heart beat into two classes (abnormal and normal class) by the extraction of their parameters using various Multi-Layered Perceptron Neural Classifiers (MLPNNs). Principal Component Analysis (PCA) is used to reduce dimensions of input features and improve the performance of the neural classifiers. This algorithm is tested on Apnea-ECG database from the universal MIT PhysioNet. As it will be shown later, the proposed algorithm allows to achieve high classification performances, describes both by sensitivity, specificity and the rate of correct classification parameters.
Benali Medjahed Oussama, Bachir M`Hamed Saadi and Hadj Slimane Zine-Eddine, 2015. Extracting Features from ECG and Respiratory Signals for Automatic Supervised Classification of Heartbeat Using Neural Networks. Asian Journal of Information Technology, 14: 53-59.