Review on spiking neural network-based ECG classification methods for low-power environments.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Electrocardiogram (ECG) signal analysis is crucial for early heart disease detection and treatment.
• Traditional methods relied on handcrafted features and machine learning, while deep neural networks (DNNs) offer high performance but are computationally intensive.
• The need for efficient, low-power solutions for real-time arrhythmia detection, especially in wearable devices, is growing.

Purpose of the Study:

• To review studies on Spiking Neural Network (SNN) based ECG classification, focusing on low-power applications.
• To provide an overview of conventional and DNN-based ECG classification methods as a foundation.
• To highlight the potential of SNNs for efficient and accurate arrhythmia detection.

Main Methods:

• Review of existing literature on ECG arrhythmia classification.
• Focus on SNN-based approaches and their comparison with DNNs.
• Analysis of computational complexity and power consumption of different methods.

Main Results:

• SNN-based ECG classification achieves accuracy comparable to DNN-based methods.
• SNNs demonstrate significantly lower computational complexity and power consumption than DNNs.
• Integration with neuromorphic hardware further enhances the ultra-low-power performance of SNNs.

Conclusions:

• SNNs present a viable alternative to DNNs for ECG arrhythmia classification, particularly in resource-constrained environments.
• The ultra-low-power capabilities of SNNs, especially with neuromorphic hardware, enable applications in lightweight wearable devices.
• This review provides valuable insights for researchers and engineers developing next-generation ECG classification systems.