On one sheep micro-SEEG shaft, high-frequency bands had lower participation ratios than 1–40 Hz after common median reference

This analysis tested whether high-frequency signals provide more independent channels than 1–40 Hz local field potentials on a high-density micro-SEEG shaft. We analyzed one 25-s sheep recording (OpenNeuro ds004819 v1.0.0, run-03; 64 contacts at 120 μm pitch; 20 spinal pulses at 0.2 mA), excluding 0.05 s before through 0.45 s after each pulse. In the 9.75 s retained, three contacts were excluded by a robust log-variance rule, and a common median reference was applied to the 61 remaining contacts. Participation ratios (PRs) of channel correlation matrices were 6.37 for 1–40 Hz, 3.25 for 70–150 Hz, and 2.93 for 300–3000 Hz. The 300–3000 Hz minus 1–40 Hz PR difference was −3.44 (95% paired window-bootstrap CI −3.93 to −1.54; 500 resamples). A post-hoc analysis excluding 20 contacts with listed impedance below 200 ohm gave the same direction (−3.71, 95% CI −3.76 to −1.76; 41 contacts). Thus, the hypothesis that the high-frequency band had a greater PR was not supported in either contact set. Contacts more than 2 mm apart had median correlations of −0.09 at 1–40 Hz and 0.16 at 300–3000 Hz; the source of the shared high-frequency component was not identified. One animal, one shaft, one short run, no task, unknown anaesthesia state, incomplete planned analyses, and post-hoc contact exclusion limit interpretation. The result does not establish independent neural sources from the higher 1–40 Hz PR after common median referencing.

micro-SEEG · intracranial EEG · participation ratio · common median reference · high-frequency activity · local field potential · channel correlation · sheep · brain-computer interface