Key result
Novel ECG deep learning model classifies arrhythmia and heart failure with ~98% accuracy.
Why the study?
Visual inspection and analysis of ECG signals for cardiac arrhythmias are difficult and time-consuming, necessitating an automated system to assist diagnosis.
Does a CNN-LSTM hybrid deep learning model using Morse-based time-frequency representation accurately detect cardiac arrhythmias from ECG data?
Does a CNN-LSTM hybrid deep learning model using Morse-based time-frequency representation accurately detect cardiac arrhythmias from ECG data?
A CNN-LSTM hybrid deep learning model using Morse wavelet-transformed ECG images can accurately classify normal sinus rhythm, congestive heart failure, and arrhythmias.
An irregular heartbeat caused by anomalies in the electrical conduction of the heart muscle is known as a cardiac arrhythmia. ECG devices are used to noninvasively monitor and diagnose cardiac arrhythmias in clinical settings. Visual inspection and analysis of ECG signals are difficult and time-consuming due to their dynamic nature and the abundance of complex information they convey. Therefore, an automated system that can distinguish between abnormal and normal ECG signals is required to assist doctors in quickly and reliably identifying cardiac arrhythmias. The primary objective of this work is to use transfer learning algorithms and a Morse-based time-frequency representation to improve the diagnosis of cardiac arrhythmias from ECG data. A CNN—LSTM hybrid deep learning model was shown to identify cardiac arrhythmias using CWT images of ECG data. The suggested method's accuracy for ARR, CHF, and NSR was 98.0%, 96.0%, and 98.0%, respectively.
No takes yet. Share an insight, caveat, or question.
Gurudiwan et al. (2025) studied Cardiac Arrhythmias (n=162). Deep learning models (CNN-LSTM / ResNet50) using Morse wavelet CWT images vs. AlexNet or baseline models was evaluated on Classification accuracy for arrhythmia (ARR), congestive heart failure (CHF), and normal sinus rhythm (NSR). A deep learning model using continuous wavelet transform images of ECG data achieved classification accuracies of 98.0% for arrhythmia, 96.0% for congestive heart failure, and 98.0% for normal sinus rhythm.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: