Métodos de extracción de características en el ECG: Análisis comparativo

Abstract

ECG feature extraction plays a significant role in diagnosing most of the cardiac diseases. In this paper, a com- parison between three ECG feature extraction methods is presented. The methods are: Linear Principal Components Analysis (PCA), Discrete Cosine Transformation (DCT) and Kernel Principal Components Analysis (KPCA). A Multilayer Perceptron is used as a gold classifier combined with the three methods before. The beats used for training and validation of the three combinations are extracted from twelve MIT - BIH Arrhythmia Database records. The performance of the three combinations is discussed and a simple execution time is evaluated. © 2013 Springer.