Scholarlink

<p>The diagnosis of illnesses involving myopathy and neuropathy largely depends on the accurate interpretation of surface electromyograms (sEMGs), which represent the electrical activity in muscles. However, sEMG recordings are often distorted by artifacts such as additive white Gaussian noise (AWGN), baseline wander (BW), electrode motion (EM), powerline interference (PLI), and ECG interference, compromising diagnostic accuracy. To ensure precise sEMG signal analysis and facilitate automated disease detection, this study proposes an improved, robust, and optimal adaptive noise cancellation (ANC) methodology. This enhanced ANC employs an efficient arithmetic optimization algorithm (AOA) to dynamically adjust filter coefficients, reducing the root mean square error (RMSE) between the target and filtered signals. Experiments conducted on real sEMG signals corrupted with AWGN, BW, EM, PLI, and ECG interference demonstrate the effectiveness of the AOA-based ANC. Rigorous experimental analysis shows that the proposed AOA-based ANC outperforms the bat algorithm (BATA), cuckoo search algorithm (CSA), particle swarm optimization (PSO), and starfish optimization algorithm (SFOA)-based ANCs, as well as other contemporary sEMG noise removal techniques, based on several standard evaluation metrics, including mean square error (MSE), correlation coefficient (CC), mean difference (MD), signal-to-noise ratio (SNR), maximum error (ME), and normalized root mean square error (NRMSE) under the given experimental conditions. To assess the practical use of the proposed ANC system, a multiclass EMG classification framework using superlet transform (SLT)-based spectrograms and the DenseNet-201 convolutional neural network (CNN) is developed to classify sEMG signals into thumb up (TU), pointing index (PI), and wrist extension with a closed hand (WEWCH) movements.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing EMG signals: an approach with optimized adaptive filtering

  • 期刊论文

  • 发表日期：

    2026年4月29日

查看全文
  • Madhava Rao Alla,
  • Chandan Nayak

摘要

The diagnosis of illnesses involving myopathy and neuropathy largely depends on the accurate interpretation of surface electromyograms (sEMGs), which represent the electrical activity in muscles. However, sEMG recordings are often distorted by artifacts such as additive white Gaussian noise (AWGN), baseline wander (BW), electrode motion (EM), powerline interference (PLI), and ECG interference, compromising diagnostic accuracy. To ensure precise sEMG signal analysis and facilitate automated disease detection, this study proposes an improved, robust, and optimal adaptive noise cancellation (ANC) methodology. This enhanced ANC employs an efficient arithmetic optimization algorithm (AOA) to dynamically adjust filter coefficients, reducing the root mean square error (RMSE) between the target and filtered signals. Experiments conducted on real sEMG signals corrupted with AWGN, BW, EM, PLI, and ECG interference demonstrate the effectiveness of the AOA-based ANC. Rigorous experimental analysis shows that the proposed AOA-based ANC outperforms the bat algorithm (BATA), cuckoo search algorithm (CSA), particle swarm optimization (PSO), and starfish optimization algorithm (SFOA)-based ANCs, as well as other contemporary sEMG noise removal techniques, based on several standard evaluation metrics, including mean square error (MSE), correlation coefficient (CC), mean difference (MD), signal-to-noise ratio (SNR), maximum error (ME), and normalized root mean square error (NRMSE) under the given experimental conditions. To assess the practical use of the proposed ANC system, a multiclass EMG classification framework using superlet transform (SLT)-based spectrograms and the DenseNet-201 convolutional neural network (CNN) is developed to classify sEMG signals into thumb up (TU), pointing index (PI), and wrist extension with a closed hand (WEWCH) movements.