Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MECG-E: Mamba-based ECG Enhancer for Baseline Wander Removal

Kuo-Hsuan Hung, Kuan-Chen Wang, Kai-Chun Liu, Wei-Lun Chen, Xugang Lu, Yu Tsao, Chii-Wann Lin

TL;DR

A novel ECG denoising model, namely Mamba-based ECG Enhancer (MECG-E), which leverages the Mamba architecture known for its fast inference and outstanding nonlinear mapping capabilities is proposed, demonstrating the model’s functionality and efficiency.

Abstract

Electrocardiogram (ECG) is an important non-invasive method for diagnosing cardiovascular disease. However, ECG signals are susceptible to noise contamination, such as electrical interference or signal wandering, which reduces diagnostic accuracy. Various ECG denoising methods have been proposed, but most existing methods yield suboptimal performance under very noisy conditions or require several steps during inference, leading to latency during online processing. In this paper, we propose a novel ECG denoising model, namely Mamba-based ECG Enhancer (MECG-E), which leverages the Mamba architecture known for its fast inference and outstanding nonlinear mapping capabilities. Experimental results indicate that MECG-E surpasses several well-known existing models across multiple metrics under different noise conditions. Additionally, MECG-E requires less inference time than state-of-the-art diffusion-based ECG denoisers, demonstrating the model's functionality and efficiency.

MECG-E: Mamba-based ECG Enhancer for Baseline Wander Removal

TL;DR

A novel ECG denoising model, namely Mamba-based ECG Enhancer (MECG-E), which leverages the Mamba architecture known for its fast inference and outstanding nonlinear mapping capabilities is proposed, demonstrating the model’s functionality and efficiency.

Abstract

Electrocardiogram (ECG) is an important non-invasive method for diagnosing cardiovascular disease. However, ECG signals are susceptible to noise contamination, such as electrical interference or signal wandering, which reduces diagnostic accuracy. Various ECG denoising methods have been proposed, but most existing methods yield suboptimal performance under very noisy conditions or require several steps during inference, leading to latency during online processing. In this paper, we propose a novel ECG denoising model, namely Mamba-based ECG Enhancer (MECG-E), which leverages the Mamba architecture known for its fast inference and outstanding nonlinear mapping capabilities. Experimental results indicate that MECG-E surpasses several well-known existing models across multiple metrics under different noise conditions. Additionally, MECG-E requires less inference time than state-of-the-art diffusion-based ECG denoisers, demonstrating the model's functionality and efficiency.
Paper Structure (14 sections, 5 equations, 5 figures, 1 table)

This paper contains 14 sections, 5 equations, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Model Architecture
  5. Encoder
  6. Time-Frequency Bi-direcional Mamba (TF-Bi-Mamba)
  7. Decoder
  8. Loss function
  9. Experiment
  10. Dataset
  11. Implementation Details
  12. Baselines and Evaluation metrics
  13. Results
  14. Conclusion

Figures (5)

  • Figure 1: The overview of the proposed MECG-E model. The dashed line represents the corresponding module. The SSM modules refer to the selective state space mechanism mentioned in Sec.\ref{['sec:mamba']}.
  • Figure 2: Evaluation metrics under different noise magnitudes. The x-axis represents the noise levels.
  • Figure 3: Results for different compression exponents. The x-axis represents compression exponents from 0.1 to 0.9 in 0.1 increments. The red star marks the setting used in this study.
  • Figure 4: Relationship between SSD and inference time under various settings. The "w.l." and the number on the graph represent the window size and hop length in STFT, respectively. The red star highlights the setting used in this study.
  • Figure 5: Reconstructed signals by the proposed MECG-E model across different noise amplitudes.