Open AccessArrhythmia Detection Using Convolutional Neural Models
Author(s) -
Jorge Torres Ruiz,
J. David Buldain Pérez,
José Ramón Beltrán Blázquez
Publication year - 2019
Publication title -
jornada de jóvenes investigadores del i3a
Language(s) - English
Resource type - Journals
ISSN - 2341-4790
DOI - 10.26754/jji-i3a.003522
Subject(s) - convolutional neural network , spectrogram , artificial intelligence , computer science , pattern recognition (psychology) , transfer of learning , task (project management) , speech recognition , engineering , systems engineering
Our main goal was studying the effectiveness of transfer learning using 2D CNNs. For this task, we generated spectrograms from ECG segments that were fed to a CNN to automatically extract features. These features are classified by a MLP into arrhythmic or normal rhythm segments, achieving 90% accuracy.