Mains harmonics exceeded 1% of high-gamma power on 30–57% of monopolar SEEG channels

This analysis measured mains-harmonic power in the 70–150 Hz high-gamma band of SEEG recordings from four public OpenNeuro datasets. It used one 30-s segment from each of 23 recordings (up to six subjects per dataset), beginning 60 s after recording start. For monopolar channels, 29.7–56.9% per dataset had estimated harmonic excess above 1%; 0–9.8% exceeded 10%. A 110 Hz non-harmonic control estimated the measure's noise floor. Adjacent-contact bipolar referencing reduced the share above 1% to 2.7–9.9%, but did not remove all harmonics. Across recordings, the share above 1% was lower with bipolar referencing in 16, higher in 4, and tied in 3 (exact sign test p = 0.012). The median excess fraction was lower in 16 of 23 recordings, a nonsignificant result (p = 0.093); most channels were near the noise floor. High-gamma channel maps changed little in most recordings (Spearman rho ≥ 0.99), while one monopolar recording had rho = 0.83. Mains frequency was assumed to be 60 Hz for two datasets whose sidecars reported n/a. The estimate counts only power above a local baseline within ±1 Hz of harmonics and therefore misses broad or modulated components. Each recording contributed only one 30-s segment; decoding, clinical mapping, and notch filtering were not tested.

SEEG · high-gamma power · mains harmonics · line noise · bipolar referencing · signal processing · OpenNeuro