A Novel Method of Segmentation and Classification for Meditation in Health Care Systems

Area of Science:

• Cognitive Science
• Neuroscience
• Physiological Research

Background:

• Electroencephalogram (EEG) is crucial for understanding neuro-physiological changes during meditation.
• EEG signal analysis faces challenges due to noise and large datasets.
• Preprocessing is essential to remove artifacts and noise from raw EEG signals.

Purpose of the Study:

• To develop an automated and precise method for EEG signal segmentation and artifact removal.
• To evaluate the effectiveness of novel signal processing techniques for classifying meditation experiences.
• To identify key EEG features indicative of meditation states.

Main Methods:

• Raw EEG signals were preprocessed using a Band-Pass Filter (BPF) for noise reduction.
• Adaptive Sliding Window with Fuzzy C Means Clustering (SW-FCM) was employed for automatic signal segmentation.
• Five features (alpha spectrum, peak frequency/amplitude, HOC, wavelet features) were extracted for analysis.
• Fuzzy Kernel Least Square Support Vector Machine (FKLSSVM) classifier was used for meditation experience classification.
• Data was sourced from the Open Brain-Computer Interface (Open BCI) platform.

Main Results:

• The SW-FCM method demonstrated precise and automatic segmentation of EEG signals.
• The selected features effectively differentiated between meditation and non-meditation states.
• The FKLSSVM classifier achieved valuable performance in classifying meditation experiences.
• Comparative analysis using MATLAB validated the presented methods against diverse techniques.

Conclusions:

• The proposed method offers a robust and efficient approach for EEG signal analysis in meditation research.
• Accurate classification of meditation experiences can be achieved using advanced signal processing and machine learning.
• This technique provides a valuable tool for understanding the neuro-physiological underpinnings of meditation.