Scholarlink

<p>Electroencephalogram (EEG) signal is a useful therapeutic tool for detecting, monitoring, and treating neurological diseases. The clinical value of this signal is typically harmed by the signal contaminations or artifacts, particularly biological artifacts such as "Electrooculogram (EOG), Electrocardiogram (ECG), and Electromyogram (EMG)". Generally, EEG artifact reduction techniques have been assumed as a critical first step in EEG analysis. Yet, in order to efficiently exploit the data, the related problems of EEG artifact reduction procedures need to be carefully labeled. Various methods for suppressing these artifacts have been developed, and most of these have shown significant improvements over uncorrected signals. The strategic performance of these algorithms, on the other hand, has not been adequately evaluated. Therefore, the research work presents a new deep learning-based EEG artifacts removal model by gathering the dataset from the standard publically available database. The development of the heuristic-based Cascaded Convolutional Neural Network (CCNN) for artifact removal is the fundamental purpose of the proposed method. CCNN is a novel deep learning architecture comprised of multiple CNNs. Here, the parameter optimization of the CCNN is performed by the hybrid Bird Swarm-based Harris Hawks Optimization (BS-HHO). The performance of the CCNN is validated by assigning a multi-objective fitness function with signal performance metrics like “Peak Signal-To-Noise Ratio (PSNR) and Root-Mean-Square Error (RMSE)”. The suggested technique has been compared to existing automated artifact removal algorithms on different artifacts datasets. Its improved performance has been demonstrated using a variety of performance measures.</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

Heuristic-based cascaded convolutional neural network for enhanced artifacts removal from EEG signal

  • 期刊论文

  • 发表日期：

    2025年7月25日

查看全文
  • Mathe Mariyadasu,
  • Mididoddi Padmaja,
  • Battula Tirumala Krishna

摘要

Electroencephalogram (EEG) signal is a useful therapeutic tool for detecting, monitoring, and treating neurological diseases. The clinical value of this signal is typically harmed by the signal contaminations or artifacts, particularly biological artifacts such as "Electrooculogram (EOG), Electrocardiogram (ECG), and Electromyogram (EMG)". Generally, EEG artifact reduction techniques have been assumed as a critical first step in EEG analysis. Yet, in order to efficiently exploit the data, the related problems of EEG artifact reduction procedures need to be carefully labeled. Various methods for suppressing these artifacts have been developed, and most of these have shown significant improvements over uncorrected signals. The strategic performance of these algorithms, on the other hand, has not been adequately evaluated. Therefore, the research work presents a new deep learning-based EEG artifacts removal model by gathering the dataset from the standard publically available database. The development of the heuristic-based Cascaded Convolutional Neural Network (CCNN) for artifact removal is the fundamental purpose of the proposed method. CCNN is a novel deep learning architecture comprised of multiple CNNs. Here, the parameter optimization of the CCNN is performed by the hybrid Bird Swarm-based Harris Hawks Optimization (BS-HHO). The performance of the CCNN is validated by assigning a multi-objective fitness function with signal performance metrics like “Peak Signal-To-Noise Ratio (PSNR) and Root-Mean-Square Error (RMSE)”. The suggested technique has been compared to existing automated artifact removal algorithms on different artifacts datasets. Its improved performance has been demonstrated using a variety of performance measures.