Classification of ECG Signals Based on Hilbert-Huang Transform and 1D Convolution Neural Network

Abstract

Arrhythmia is a common phenomenon in cardiovascular diseases and its accurate diagnosis typically relies on a thorough analysis of electrocardiogram (ECG). Therefore, accurate identification and classification of ECG signals play a crucial role in the effective treatment of cardiac diseases. In this study, we propose a deep learning model based on Hilbert-Huang Transform (HHT) and one-dimensional convolutional neural network (1D-CNN), aiming to enhance the accuracy of ECG signal classification. Specifically, we first perform time-frequency analysis and feature extraction of the signal using the Hilbert-Huang Transform, after which we further extract features and classify the signal using 1D Convolutional Neural Network. The experimental results show that the model proposed in this paper performs well in classifying four types of ECG signals, with average classification accuracies of 97.73%, 99.16%, 99.50%, and 99.88%, respectively. This not only proves the effectiveness of our proposed method but also provides important technical support for the diagnosis and treatment of cardiovascular diseases, which has far-reaching clinical application value.