Lateralization in Scalp EEG Brain Connectivity During Hand Motor Imagery Can Improve Task Classification for Brain-Computer Interfaces

Objective: To evaluate brain connectivity reorganization during motor imagery (MI) tasks and assess the utility of EEG-based networks for MI classification compared to 8-13-Hz power spectrum of selected EEG channels, which are commonly used in MI decoders. Approach: We applied phase locking value (PLV), cross-correlation (CC), weighted phase lag index (wPLI), and Granger causality (GC) to left/right-hand MI task data from the BCI Competition IV 2a (BCI-IV-2a) and PhysioNet Motor Imagery (PHYS-MI) datasets, and we identified the connectivity measure with the highest predictive value for MI decoding after excluding subjects who performed at the chance level. We used Random Forest models for task classification, identified key nodes and edges, and compared results against decoders based on EEG single-channel power (sc-EEG). Main results: PLV outperformed other connectivity measures at MI decoding, with accuracy comparable to single channel EEG power (65.3±11.0% vs. 61.3±11.0% and 58.4±9.9% vs. 58.6±15.7%, mean±S.D. across subjects for BCI-IV-2a and PHYS-MI, respectively). CC also improved over wPLI and GC (one-way ANOVA test with Tukey’s HSD post-hoc test, P -value P <0.05), with CC peaking around 0-ms-lag, which suggests zero-lag synchrony between EEG nodes. Moderate correlation ( R ² =0.62 and 0.40 for BCI-IV-2a and PHYS-MI, respectively) was found between the mean difference, Δ _EVC , in eigenvector centrality of the nodes of the PLV-based network in left- vs. right-hand MI and the GINI importance score of the single-channel power values. Also, while the PLV-based network topology remained stable over time, a small set of connections (7.8±4.5% and 3.1±2.5% of edges) lateralized to the hemisphere contralateral to the movement altered considerably and enhanced classification accuracy by 6.7±5.6% and 16.3±7.5% across subjects. Significance: PLV-based EEG networks are as effective as traditional band-limited EEG power at decoding MI. These findings underscore the importance of connectivity reorganization in MI, with task-specific pathways enhancing decoding without disrupting overall network topology and offer insights for improving brain-computer interface systems.