HBS: a novel biometric feature based on heartbeat morphology

Area of Science:

• Biometrics
• Cardiovascular Technology
• Signal Processing

Background:

• Electrocardiogram (ECG) signals are unique to individuals.
• ECG-based biometrics offers a non-invasive identification method.
• Existing ECG biometric methods face challenges with signal variability and efficiency.

Purpose of the Study:

• To propose a new feature, heartbeat shape (HBS), for robust ECG-based biometrics.
• To evaluate the effectiveness of HBS for both identification and verification tasks.
• To assess the efficiency of HBS in telecardiology contexts.

Main Methods:

• HBS feature extraction involves heartbeat segmentation, resampling, normalization, error reduction, and shift-invariant transformation.
• Templates are constructed using a few consecutive heartbeats for efficient processing.
• The HBS feature was tested on two public ECG databases, including one with cardiac irregularities.

Main Results:

• High identification accuracy was achieved: 98% on one database and 99.85% on another.
• Low equal error rates for verification were reported: 1.88% and 0.38%, respectively.
• Effective performance was demonstrated even with subjects experiencing cardiac irregularities like arrhythmia.

Conclusions:

• The proposed heartbeat shape (HBS) feature provides a highly accurate and efficient method for ECG-based biometrics.
• HBS demonstrates robustness, performing well even with subjects having cardiac irregularities.
• The significant compression of ECG signals by HBS makes it suitable for efficient telecardiology communication and data access.