No consistent anterior-posterior sigma peak gradient was detected in pediatric sleep SEEG

We tested whether sigma (spindle-band) spectral peak frequency is higher at posterior than anterior neocortical SEEG pairs during sleep. We used 120 s from 18 subjects in OpenNeuro ds005398 and wrote and hashed the analysis plan before reading signals. For each bipolar pair, we removed a fitted aperiodic line from the log spectrum and found the residual peak in 9-16 Hz. Fifteen patients had at least 10 valid non-SOZ neocortical pairs and a y range of at least 30 mm. The mean within-patient Spearman rho between MNI y and peak frequency was -0.064 (95% CI -0.235 to +0.104; one-sided exact sign-flip p = 0.24; 9 of 15 negative). The primary hypothesis was not supported. Secondary tests (partial rho for z and |x|, SOZ pairs included, parietal minus frontal median, and spectral centroid) were also not significant after Holm correction. The parietal-minus-frontal difference was +0.46 Hz (CI -0.10 to +1.31) in only 6 patients. Limits include children with epilepsy, clinical coverage with few parietal pairs, unscored sleep stage (median delta fraction 0.71), short segments, and 15% of pairs with a peak at the 16 Hz band edge excluded by the plan rule. The package includes the plan, code, per-pair table, PSDs, and byte-level provenance.

sleep · sigma oscillations · spindles · SEEG · spectral peak · pediatric epilepsy · OpenNeuro