Arrhythmia Classification with Single-Channel Features Extracted from "A Large-Scale 12-Lead ECG Database for...

Area of Science:

• Biomedical Engineering
• Cardiology
• Signal Processing

Background:

• Automated arrhythmia classification is crucial for cardiac care.
• Traditional electrocardiogram (ECG) analysis often relies on multiple leads, increasing complexity.
• Developing efficient and accurate classification methods is essential for widespread clinical adoption.

Purpose of the Study:

• To evaluate the effectiveness of classical and modern ECG features from a single lead (Lead II) for automated arrhythmia classification.
• To compare the performance of single-lead ECG analysis with multi-lead approaches.
• To assess the efficiency gains of using a single ECG lead.

Main Methods:

• Utilized the Large Scale 12-Lead Electrocardiogram Database for Arrhythmia Study.
• Extracted classical morphological features (e.g., QRS duration, QT interval) and advanced time-, frequency-, and nonlinear-domain descriptors from a single ECG lead.
• Employed feature selection techniques (e.g., MRMR) and classification algorithms for four and eight arrhythmia categories.

Main Results:

• Achieved 94.2% accuracy in a four-class task using 15 MRMR-selected features.
• Attained 69% accuracy in an eight-class task with 29-39 features.
• Demonstrated a ~12-fold reduction in preprocessing, storage, and classification time compared to 12-lead methods.

Conclusions:

• Advanced ECG descriptors from a single lead can achieve high accuracy in arrhythmia classification.
• Single-lead ECG analysis offers significant efficiency advantages, making it practical for scalable clinical applications.
• This approach holds promise for improving diagnostic capabilities in resource-limited settings.