Advanced ECG Signal Classification of Cardiac Conditions Via Convolutional Neural Networks

The early detection of cardiac conditions depends on ECG signal classification. ECG signals have been classed very well with the development of machine learning, particularly with deep learning. Advanced ECG signal classification using Convolutional Neural Networks (CNN), a powerful deep learning architecture, is presented in this paper. In this context, CNNs can provide automatic pattern extraction of intricate patterns from ECG wave forms for the purposes of increased accuracy in detecting different cardiac abnormalities. Using multiple layers of convolutions, pooling, and fully connected networks, the proposed approach classifies these complex cardiac conditions including arrhythmia, atrial fibrillation and congestive heart failure. An important aspect of the proposed method is that it can perform end to end classification directly from the raw ECG signal input to the classification output, avoiding unnecessary feature engineering. Further improvement in the performance is realized by removing noise through maximal overlap discrete wavelet packet transform (MODWPT) inside feature extraction process. An ECG dataset of decent size was used to train and validate the model, resulting in high accuracy and reliability of separation between normal and abnormal heart conditions. This work demonstrates the effectiveness of CNNs in medical signal processing towards real time and automatic ECG monitoring systems.