ECG Signal Classification using Smoothed Pseudo Wigner-Ville Distribution

The electrocardiogram (ECG) contains the information required for identifying various cardiovascular diseases or abnormalities. Early identification of abnormalities will be immensely useful in the treatment of the abnormality. In this work, the ECG signals transforming into time-frequency images using the smoothed pseudo Wigner-Ville distribution (SPWVD). These images are split to train and test dataand fed to a Convolutional Neural Network (CNN) model, distinguishing between normal and abnormal heartbeats. Here, we use PTB database. The proposed method based on SPWVD images achieves an accuracy of 98.96%. The method discussed in this work is compared with the methods available in the literature. The effectiveness of the SPWVD-based method in accurately identifying abnormal heartbeats from ECG signals, suggesting its potential for clinical application.