A neuromorphic approach to early arrhythmia detection.

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Artificial Intelligence

Background:

• Accurate electrocardiogram (ECG) analysis is vital for diagnosing cardiovascular diseases.
• Existing arrhythmia detection methods face challenges in capturing complex temporal dynamics.

Purpose of the Study:

• To develop and evaluate a novel bio-inspired Spiking Neural Network (SNN) for enhanced ECG arrhythmia detection.
• To leverage biological neural mechanisms, specifically leaky integrate-and-fire (LIF) neurons and spike timing-dependent plasticity (STDP), for improved classification accuracy.

Main Methods:

• ECG signal preprocessing including normalization and filtering.
• Transformation of continuous ECG signals into spike trains using rate coding.
• Implementation of an SNN model with LIF neurons and STDP for learning temporal patterns.

Main Results:

• The SNN model achieved an overall accuracy of 94.4% in arrhythmia detection.
• Sensitivity and F1-scores exceeded 0.88 across all arrhythmia classes.
• Exceptional performance in identifying left bundle branch block (LBBB) and right bundle branch block (RBBB) with F1-scores of 1.00 and 0.99, respectively.

Conclusions:

• The bio-inspired SNN effectively captures critical temporal dynamics for accurate arrhythmia classification.
• The model demonstrates high reliability in detecting various arrhythmias, including LBBB and RBBB.
• This approach shows significant potential for advancing automated ECG interpretation and clinical decision support.