Classification of meditation states through EEG: A method using discrete wavelet transform

Abstract

Meditation is a commonly adopted lifestyle practice for the associated benefits in increasing mental and emotional wellbeing. Meditation can be described as a non-physical exercise of being in a sustained state of mental relaxation and focused attention. When exiting a meditation state, be it voluntarily or otherwise externally disturbed, a transition Electroencephalogram (EEG) signal can be observed. This study proposed an extraction of EEG transition feature of the signal between state of meditation and post meditation using Discrete Wavelet Transform (DWT). Wavelet analysis retains both time and frequency information, allowing the transition feature to be easily identified and localised on the time scale. When given a continuous meditation EEG signal as input, results from DWT were applied to detect time localised transition during meditation. Results from our studies has demonstrated that the DWT feature extraction model is capable of detecting when a meditation practitioner exited meditation state at a 96.90% accuracy. This meditation transition report can be used in part as a feedback to the practitioner on their sustained meditation performance.