Heart beat classification from single-lead ECG using the synchrosqueezing transform

Area of Science:

• Cardiovascular physiology and signal processing.
• Biomedical engineering and computational biology.

Background:

• Electrocardiogram (ECG) signal processing offers vital insights into cardiac function and health.
• Single-lead ECG devices are increasingly common for ambulatory monitoring, but accurate heart rate variability (HRV) assessment relies on precise beat detection and rhythm classification, which is challenging with limited leads.
• Existing multi-lead ECG methods are not always practical, especially for applications like fetal monitoring.

Purpose of the Study:

• To develop and validate a novel adaptive non-harmonic model utilizing the synchrosqueezing transform (SST) for improved ECG pattern characterization.
• To enhance heart beat detection and classification of normal versus abnormal rhythms using single-lead ECG data.
• To demonstrate the efficacy of SST-derived features in a machine learning classifier for arrhythmia detection.

Main Methods:

• An adaptive non-harmonic model was developed to represent the heart rate signal.
• The synchrosqueezing transform (SST) was employed to characterize ECG patterns.
• A support vector machine (SVM) classifier was trained and validated using SST-derived instantaneous phase, R-peak amplitudes, and R-peak to R-peak interval durations from single-lead ECG data.
• The Massachusetts Institute of Technology-Beth Israel Hospital (MIT-BIH) arrhythmia database and Association for the Advancement of Medical Instrumentation (AAMI) beat classes were utilized for training and validation.

Main Results:

• The proposed model successfully enhanced heart beat detection and classification accuracy for normal and abnormal rhythms.
• The SST-derived features, combined with SVM, achieved sensitivities and positive predictive values comparable to established multi-lead algorithms.
• This demonstrates the potential of single-lead ECG analysis with advanced signal processing techniques.

Conclusions:

• The synchrosqueezing transform (SST) provides a powerful tool for analyzing single-lead ECG signals, enabling robust beat detection and arrhythmia classification.
• The developed adaptive non-harmonic model and SST-based feature extraction offer a viable and effective alternative to multi-lead ECG analysis for certain applications.
• This approach holds promise for improving cardiovascular health monitoring through accessible, single-lead ambulatory devices.