(INSTITUTO DE TECNOLOGIA, 2011) NETO, João Evangelista; SUAREZ-LEON, A. A.; VÁZQUEZ-SEISDEDOS, C. R; LÓPEZ-MORA, N. A; LEITE, J. C; OLIVEIRA, R. C. L.; Rui Nelson Otoni Magno
The article compares three ECG feature extraction methods (Discrete Cosine Transform - DCT, Principal Component Analysis - PCA, and Kernel PCA) for heartbeat classification using an MLP neural network. Results indicate that Kernel PCA achieves the highest accuracy (98.7%) but with longer execution times, while linear PCA is the fastest but less accurate (93%). The study uses data from the MIT-BIH Arrhythmia Database.
