Discriminating between V and N Beats from ECGs Introducing an Integrated Reduced Representation along with a Neural Network Classifier

  • Authors:

  • Vaclav Chudacek;George Georgoulas;Michal Huptych;Chrysostomos Stylios;Lenka Lhotska

  • Affiliations:

  • Dept. of Cybernetics, Czech Technical University in Prague, Prague, Czech Republic;Dept of Computer Applications in Finance and Management, TEI of Ionian Islands, Lefkas, Greece;Dept. of Cybernetics, Czech Technical University in Prague, Prague, Czech Republic;Dept. of Informatics and Telecommunications Technology, TEI of Epirus, Arta, Greece;Dept. of Cybernetics, Czech Technical University in Prague, Prague, Czech Republic

  • Venue:
  • ICANN '09 Proceedings of the 19th International Conference on Artificial Neural Networks: Part II
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

The main objective of this paper is to investigate and propose a new approach to distinguish between two classes of beats from the ECG holter recordings - the premature ventricular beats (V) and the normal ones (N). The integrated methodology consists of a specific sequence: R-peak detection, feature extraction, Principal Component Analysis dimensionality reduction and classification with a neural classifier. ECG beats of holter recordings are described using means as simple as possible resulting in a description of the QRS complex by features derived mathematically from the signal using only R-peak detection. For this research work, normal (N) and ventricular (V) beats from the well known MIT-BIH database were used to test the proposed methodology. The results are promising paving the way for the more demanding multiclass classification problem.