Preprocessing-Oriented EEG Analysis for Measuring Stress and Anxiety Alleviation curtailment by Mantra Meditation

Abstract

Electroencephalography (EEG) is becoming the key technique of researching mental states, particularly when studies focus on stress and anxiety. Regrettably, EEG recordings are routinely contaminated by numerous sources of noise, including power line noise, motion artifacts, burst noise, Electrooculography (EOG), Electromyography (EMG), and Electrocardiography (ECG), which severely affect the quality of the signal. To enhance the information, an efficient EEG preprocessing framework is developed to study the effects of Mantra Chanting on stress and anxiety. The proposed algorithm automated an artifact-classification pipeline that would identify which type of denoising can be best applied. The performance is compared for filters, such as CNF-SFFDF, Butterworth, Kernel Density Estimation with Recursive Kalman Filter, and Sliding Window Adaptive Filtering, in eight channels of EEG signal at the pre-, during, and after-meditation periods. As results indicate, Chebyshev Notch Filter (CNF) + Spatial Filtering with Frequency-Domain Filtering is the best in all cases falling through in peak PSNR (9.13%), SSIM (3.47%), and SNR (38.74%) is improved for Channel Fp1, proving to be quite efficient noise suppression and signal retention. Meditation-induced neurophysiological changes can also be measured accurately with better-quality signals, which also supports the role of mantra meditation in reducing stress. The proposed preprocessing framework offers a solid foundation when it comes to real-time mental health monitoring based on EEG analytics.