Detected spindle density was lower at seizure-onset-zone pairs, but the difference depended on spike rejection and was not independent of power

Question: in NREM-sleep SEEG, do bipolar pairs in the clinical seizure-onset zone (SOZ) have fewer sleep spindles than other pairs in the same patient? We wrote the plan (PLAN.md, with SHA-256) before we read signal. Data: OpenNeuro ds007118, ds007119 and ds007120 (1.0.0); 43 patients, 728 bipolar pairs, 160 SOZ pairs; EDF time 30-210 s of the first sleep run. We used the subject selection, pairing and EDF reader of AutoSci a3d06b3c. Detector: 11-16 Hz amplitude, mean + 3 SD, 0.5-3 s, with removal of events near 25-80 Hz spikes. Test: per-patient AUC (SOZ vs non-SOZ), sign-flip test, Holm correction over four hypotheses. Result: spindle density was lower at SOZ pairs (H1: mean AUC 0.35, 95% CI 0.28-0.41, Holm p = 0.0002; 32 of 43 patients). Relative sigma power was not lower (H2: AUC 0.56, Holm p = 0.54). After control for delta, broadband and HFA power, the effect was not significant (H3: AUC 0.43, Holm p = 0.18). The lobe-matched test was not significant (H4: 8 patients, Holm p = 0.54). Non-SOZ spiking pairs also had lower density than normal pairs (AUC 0.34, p = 0.001). Post hoc (not in the plan): without spike rejection, density was not different (AUC 0.53, p = 0.48), and SOZ pairs had more rejected events (AUC 0.66, p = 0.0002). Conclusion: the lower detected spindle density at SOZ pairs comes mainly from spike-related event removal and from power differences. It is not clear evidence that the SOZ makes fewer spindles. Limits: 180 s per patient, no epoch-level sleep scoring, one automatic detector, no visual review. The package has the code, the plan, the channel files, the EDF byte ranges with SHA-256, all results and the reproduction steps.

sleep spindles · stereo-EEG · seizure-onset zone · epilepsy · NREM sleep · spike rejection · intracranial EEG