Scholarlink

<p>Meditation has gained considerable attention for its potential benefits in promoting mental and emotional well-being. While numerous studies have investigated the neural correlates of meditation using electroencephalography (EEG), relatively little research has been dedicated to brain state classification to assess the effects of meditation in different frequency bands. EEG signals are non-stationary hence extracting correct time-frequency information is a challenging task. This paper aims to build an EEG-based meditation model for the multi-classification of the brain states into expert meditators, novice meditators, and control states. Initially, a pre-processing technique is applied to generate artifact-free data. This data is decomposed into theta, alpha, and gamma frequency bands. Morlet wavelet transforms are applied in these various bands and time-frequency images are generated. This model uses an efficient deep learning model such as a Convolutional Neural Network (CNN) for the multi-classification of brain states. The model is evaluated in different frequency bands with metrics such as accuracy, precision, f1-measure, and AUC score. Among the three bands, the highest accuracy of 99% is achieved in theta bands followed by 98% in the alpha and 91% in the gamma band. The results outperform as compared to other state of art methods. Group group-level clustering technique of Independent Component Analysis (ICA) gives deep insight into various brain states. This model serves as a base to generate real-time neuro-feedback mechanisms for meditation practitioners.</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

A meditation-based brain state classification framework: an integrated Morlet wavelet transforms and CNN approach with EEG signals

  • 期刊论文

  • 发表日期：

    2025年8月1日

查看全文
  • Soniya Usgaonkar,
  • Damodar Reddy Edla,
  • R. Ravinder Reddy

摘要

Meditation has gained considerable attention for its potential benefits in promoting mental and emotional well-being. While numerous studies have investigated the neural correlates of meditation using electroencephalography (EEG), relatively little research has been dedicated to brain state classification to assess the effects of meditation in different frequency bands. EEG signals are non-stationary hence extracting correct time-frequency information is a challenging task. This paper aims to build an EEG-based meditation model for the multi-classification of the brain states into expert meditators, novice meditators, and control states. Initially, a pre-processing technique is applied to generate artifact-free data. This data is decomposed into theta, alpha, and gamma frequency bands. Morlet wavelet transforms are applied in these various bands and time-frequency images are generated. This model uses an efficient deep learning model such as a Convolutional Neural Network (CNN) for the multi-classification of brain states. The model is evaluated in different frequency bands with metrics such as accuracy, precision, f1-measure, and AUC score. Among the three bands, the highest accuracy of 99% is achieved in theta bands followed by 98% in the alpha and 91% in the gamma band. The results outperform as compared to other state of art methods. Group group-level clustering technique of Independent Component Analysis (ICA) gives deep insight into various brain states. This model serves as a base to generate real-time neuro-feedback mechanisms for meditation practitioners.