Slow-wave coupling retains seizure-onset-zone signals after mutual control, but little remains after band-power control

We tested if the higher slow-oscillation (SO, 0.5-1.25 Hz) to sigma (12-16 Hz) coupling in clinical seizure-onset-zone (SOZ) pairs (6cfd4e54, mean AUC 0.67) comes only from broadband transients that lock to slow waves. We used the archived per-pair features of 6cfd4e54 (43 patients, 728 depth bipolar pairs, 160 SOZ pairs, OpenNeuro ds007118/9/20, 180 s NREM). We wrote and hashed the plan before we computed new statistics. In each patient, we removed the rank effect of one coupling z-score from the other and computed the SOZ AUC of the residual. Sign-flip tests over patients, Holm correction over 2 tests. Result: SO-sigma coupling kept a SOZ signal after control for SO-HFA (80-300 Hz) coupling (mean AUC 0.601, 95% CI 0.527-0.675, 29/43 patients, Holm p 0.023). SO-HFA coupling also kept a SOZ signal after control for SO-sigma coupling (0.591, CI 0.517-0.664, Holm p 0.023). In non-SOZ pairs, the two couplings had only a moderate rank correlation (mean rho 0.31). When we also controlled sigma and HFA power, little stayed (SO-sigma 0.560, p 0.051; SO-HFA 0.539, p 0.24). The SO-sigma residual was clearer in the external cohorts (n=28, 0.62) than in the Detroit cohorts (n=15, 0.56, CI includes 0.5). Thus SO-HFA coupling alone does not explain the SOZ excess of SO-sigma coupling, but most of the signal is shared with band power. Limits: archived features only, no spike or spindle detection, noisy 50-surrogate z-scores make the control incomplete, clinical SOZ labels.

slow oscillation · sigma coupling · high-frequency activity · seizure-onset zone · stereoelectroencephalography · epilepsy · rank regression · OpenNeuro