Recorded-reference SEEG correlations persisted beyond 60 mm, while bipolar data did not support a shorter high-frequency length constant

Question: how fast does SEEG signal correlation decrease with contact distance, by band and reference? Data: OpenNeuro ds005670 v1.0.0, 2 resting patients, partial EDF byte ranges (39.6 s and 34.7 s), 211 and 137 contacts. Method: Pearson r of band-limited signals (1-4, 4-8, 8-13, 13-30, 30-45, 70-150 Hz) for all pairs, against MNI distance, with 3 references: recorded contact pair (M), adjacent bipolar (B, pairs with a shared contact removed), common average (C). Exponential fit and shaft-cluster bootstrap. The plan was written before the analysis. Results: H2 was supported. Mean r for pairs > 30 mm was higher for M than for B in all 6 bands in both patients (lowest 95% CI bound +0.13). With M, mean r at > 60 mm was 0.17-0.67, also at 70-150 Hz. With B, |mean r| was 0.061 or less in all distance bins. H1 (bipolar length constant shorter for 70-150 Hz than for 1-4 Hz) was not supported. The bipolar fit went to its bounds and the CIs include 0. With C, length constants were 13-21 mm for 1-45 Hz and 26-30 mm for 70-150 Hz; the 70-150 Hz value was not stable between halves. Post hoc: same-shaft bipolar pairs at 7-10.5 mm had small negative r (-0.03 to -0.12) in beta, low gamma and 70-150 Hz. Limits: 2 patients, less than 40 s each, no spike rejection, MNI distances, no tissue labels.

SEEG · intracranial EEG · signal correlation · reference · distance decay · high-frequency activity · bipolar · common average · OpenNeuro