Bipolar SEEG slow-oscillation phase transfer weakens with distance during NREM sleep

We tested if the slow-oscillation (SO, 0.5-2 Hz) phase at one SEEG site can predict the SO phase at a second site during NREM sleep. This is necessary for closed-loop stimulation that reads at one site and stimulates at another. Data: OpenNeuro ds007118/ds007119/ds007120, 120 s per subject. We detected up-phase events on a trigger unit, calibrated a phase offset on the first half, and measured the hit rate (phase error < 45 deg; chance 0.25) on the second half. Bipolar hit rates were 0.44 at 2 contact steps and 0.30 at 9 or more steps (surrogate 0.25). The pre-registered test with fsaverage coordinates had only 6 subjects because electrodes.tsv omits most contacts (median rho -0.22, p = 0.078, not significant). A documented deviation that uses same-shaft contact steps (66 subjects) showed a decrease with distance (median rho -0.18, p = 1.5e-5). Monopolar signals gave high hit rates (0.96 at 1 step), which is probably caused by the common reference. Conclusion: with local (bipolar) signals, phase-locked stimulation must read the phase at the target site or very near it. Limits: offline zero-phase filter, short segments, no sleep scoring.

slow oscillation · stereo-EEG · SEEG · NREM sleep · phase transfer · closed-loop stimulation · bipolar signals · monopolar signals · volume conduction