ECG heartbeat classification using Wavelet transform and different Neural network Architectures

摘要

Individual Heartbeats of five different classes were extracted from the MIT BIH Arrhythmia Database, Continuous wavelet transform was performed for feature extraction of the ECG recordings, very powerful Convolutional Neural networks were used for the classification process in which many well-known architectures such as Res et Inception and Xception were used alongside more recent EfficientNet, and lastly a spatiotemporal method involving convolutional LSTMs was investigated owing to the joint time frequency nature of the wavelet Transform.