Scholarlink

Electroencephalography (EEG) data are affected by unwanted signals and noises from different sources. Diagnoses and analyses of human brain illnesses require a strong neurological signal. Therefore, detecting the EEG artifacts is a crucial step in order to facilitate the elimination process. This research paper proposes a new approach based on EEG signal analysis to automatically detect and identify EEG artifacts related to eyebrows, blinks, and head movements. The approach evaluates the use of a set of machine learning techniques to classify artifacts relating to a set of EEG data. A fusion of three types of features is extracted in this work. Finally, an evaluation phase is introduced involving a variety of evaluation measures, namely accuracy, precision, recall/sensitivity, specificity, and f-score. The most relevant captured results are presented as follows: an accuracy of 87.50%, precision of 0.93, recall/sensitivity of 0.87, specificity of 0.96, and f-score of 0.88. These results confirm that our proposed approach can be effective for identifying artifacts related to eyebrow, blink, and head movement artifacts.

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

Eyebrow, Blink and Head Movement Artifacts Detection from EEG Signals Using Machine Learning Techniques

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Rahma Mili,
  • Rania Khaskhoussy,
  • Ahmed Maalel,
  • Bassem Bouaziz,
  • Faiez Gargouri

摘要

Electroencephalography (EEG) data are affected by unwanted signals and noises from different sources. Diagnoses and analyses of human brain illnesses require a strong neurological signal. Therefore, detecting the EEG artifacts is a crucial step in order to facilitate the elimination process. This research paper proposes a new approach based on EEG signal analysis to automatically detect and identify EEG artifacts related to eyebrows, blinks, and head movements. The approach evaluates the use of a set of machine learning techniques to classify artifacts relating to a set of EEG data. A fusion of three types of features is extracted in this work. Finally, an evaluation phase is introduced involving a variety of evaluation measures, namely accuracy, precision, recall/sensitivity, specificity, and f-score. The most relevant captured results are presented as follows: an accuracy of 87.50%, precision of 0.93, recall/sensitivity of 0.87, specificity of 0.96, and f-score of 0.88. These results confirm that our proposed approach can be effective for identifying artifacts related to eyebrow, blink, and head movement artifacts.