Slow-oscillation phase locking of 11-16 Hz amplitude at normal SEEG pairs is not specific to spindles

This study used the NREM-sleep SEEG of 40 patients from OpenNeuro ds007118, ds007119 and ds007120. It used 116 s of signal per patient and 408 bipolar pairs with two normal contacts. The plan was pre-registered. The amplitude of the 11-16 Hz (sigma) band was locked to the phase of the local 0.5-1.25 Hz slow oscillation (mean patient z against a circular-shift null 0.93, 95% CI 0.56 to 1.34, Holm p = 1e-4). The preferred phase of sigma was near the phase where 70-150 Hz activity was maximal (the local up-state) (mean cos 0.42, CI 0.29 to 0.55, Holm p = 1e-4). But the slow-oscillation phase of detected spindle events was not near the up-state (mean cos 0.006, p = 0.89, 27 patients). A post-hoc check showed that 20-30 Hz amplitude was locked as strongly as sigma, and at the same phase (difference in z -0.14, p = 0.24). The sigma/20-30 Hz ratio had only weak locking, with no common phase. Thus the sigma locking agrees with a broadband increase of power in the up-state. These short segments do not show spindle-specific slow-oscillation nesting at local bipolar pairs. Pairs in the seizure-onset zone had stronger sigma locking than normal pairs (AUC 0.66, p = 1.5e-4). This agrees with an effect of interictal spikes. The package contains the code, the per-pair values, the EDF byte ranges with SHA-256, and the reproduction steps.

sleep · slow oscillations · sleep spindles · SEEG · phase-amplitude coupling · high-frequency activity · epilepsy