Classification of ECG Signals of Heart Beats using TF-TS LSTM with Augmented Fuzzy Recurrence Eigenvalues

Automated classification of physiological signals into a condition of either healthy control or disorder is useful for diagnosis and early prediction of diseases. This paper presents the extraction of the largest eigenvalues of convolutional fuzzy recurrence plots of ECG signals. This type of feature can enhance the discriminating power of long short-term memory networks for classifying ECG-based heartbeat conditions. Experimental results show the augmentation of the largest eigenvalues with time-frequency and time-space features can improve the classification performance of the deep-learning network.