SNST: New EEG measure reveals hidden amplitude coupling in brain signals

Traditional EEG connectivity analysis relies heavily on phase synchronization measures like Phase Lag Index (PLI), which are susceptible to volume conduction artifacts and ignore amplitude-domain coupling. A new paper from researchers at Bangladesh University of Engineering and Technology introduces the Spatial Neighboring Scattering Transform (SNST), extending wavelet scattering to multichannel settings. SNST produces two descriptors: a first-order descriptor capturing amplitude-envelope coupling between neighboring electrodes, and a second-order descriptor revealing how this coupling is modulated by slow rhythms, uncovering cross-frequency amplitude-modulation structures.

Validated on the BCI Competition IV-2a motor imagery dataset with a bias-corrected, false-discovery-rate-controlled pipeline, SNST found statistically significant amplitude coupling within a central-parietal electrode neighborhood consistently across all subjects and both imagery conditions. In contrast, PLI and weighted PLI, computed under identical correction, identified negligible significant coupling with zero overlap. This demonstrates that amplitude envelope coupling constitutes a distinct, previously undetected connectivity signal. SNST is applicable to any multichannel EEG analysis where amplitude-domain inter-regional dependence is of interest, offering a systematic method to recover amplitude and cross-frequency coupling.