Bipolar SEEG power increases with inter-contact distance in two independent RAM task datasets

Bipolar SEEG pairs have different inter-contact distances, and atlases and classifiers pool these pairs. We tested if bipolar power changes with this distance inside each patient. We used archived per-pair Welch spectra (package c3188d78; 60 s from the first word event, 1-45 Hz). We calculated the distance between the two contacts of each pair from the OpenNeuro MNI152NLin6ASym electrodes.tsv files. We wrote and hashed the plan before the analysis. For each patient, we calculated the Spearman rho of band power with distance. Then we did a sign-flip test over patients with Holm correction across 7 outcomes. Discovery set: ds004789 FR1, 182 patients, 17227 pairs. Replication set: ds004809 catFR1, 71 different patients, 9151 pairs. In the discovery set, power increased with distance in all six bands (broadband rho +0.091, 95% CI +0.061 to +0.120; band rho +0.065 to +0.101; Holm p 0.00035 each; rho > 0 in 65-74% of patients). The replication set gave the same result (broadband rho +0.087, CI +0.052 to +0.121; all six bands Holm p 0.0039 or less). When we controlled for the MNI position of the pair, log10 power increased by 0.12 to 0.20 per doubling of distance (1.2 to 2.0 dB). With adjacent same-shaft pairs only, the effect was smaller but present (broadband rho +0.068 and +0.053). The 2-40 Hz spectral slope did not change reliably with distance (rho -0.022, Holm p 0.14 in discovery). Limits: MNI distances come from a nonlinear warp and are not the physical contact spacing. The data are task segments, not rest. Distance can co-vary with tissue type. Thus, studies that pool bipolar power across electrode models or sites must control for inter-contact distance.

stereo-EEG · bipolar power · inter-contact distance · intracranial EEG · spectral power · OpenNeuro · RAM · Welch spectrum