Automated classification of listening phases during mantra meditation from EEG using empirical Fourier-Bessel decomposition

Mantra meditation (MM) is associated with measurable changes in neural oscillatory activity and cognitive states. However, automated discrimination of distinct neural phases during MM listening remains challenging because electroencephalogram (EEG) signals are non-stationary and highly variable across subjects. This study proposes an empirical Fourier-Bessel decomposition (EFBD)-based time–frequency representation (TFR) framework to classify three consecutive phases of a Rudram mantra (RM) listening session: Ganapati atharvashirsha (pre-listening), Namakam (active listening), and Chamakam (post-listening). Multichannel EEG signals are recorded from 28 subjects during a single RM session and segmented into 5 s epochs. Each epoch is decomposed into oscillatory mode components using EFBD, and the Hilbert transform is applied to construct TFRs. Ten TFR-based features are extracted, and the five most discriminative are selected using the minimum redundancy maximum relevance (mRMR) algorithm. Five machine learning classifiers, including ensemble boosting (EB), are evaluated under subject-dependent (SD) and leave-one-subject-out (LOSO) protocols. The proposed framework achieves accuracies of 77.5% and 74.2% under SD and LOSO, respectively, using the EB classifier. The active listening phase achieves the highest per-class accuracy of 90.9% under the SD protocol, suggesting a more consistent EEG pattern during active mantra listening. This is further supported by spatial analysis identifying the frontal and temporal regions as the most discriminative brain areas. EEG rhythm energy analysis also reveals increased theta and alpha activities during active listening, consistent with established EEG markers of meditative focus. These results demonstrate that the proposed framework provides an effective and physiologically interpretable approach for characterising meditation-related brain states during RM listening.