Bipolar-pair counts identified the clinical SOZ side more often than interictal feature rules in 36 bilaterally implanted patients

We compared a rule that selects the hemisphere with more implanted bipolar pairs with interictal spike rate, 80–150 Hz power, and line-length rules in 36 bilaterally implanted SEEG patients with a one-sided clinical seizure-onset zone (SOZ). The archived 120-s feature tables covered 15 patients from OpenNeuro ds003876 and 21 from ds004100. Pair counts identified the clinical SOZ side in 33/36 patients (0.92; exact one-sided binomial p=1.1×10⁻⁷). Higher mean within-patient z scores identified it in 26/36 for spike rate, 13/35 for 80–150 Hz power, and 19/36 for line length. Each feature rule was less accurate than the count rule (exact McNemar tests with Holm correction: p=0.039, 6.3×10⁻⁵, and 0.0026, respectively). The side of the highest-valued pair did not exceed the hit count expected from side-specific pair fractions (Holm-adjusted p=0.63–0.98). These findings show that implant layout carried most of the information about the clinical SOZ side in these cohorts; they do not show that pair counts identify the true epileptogenic side. The clinical SOZ was determined on the same implant, only three patients had the SOZ on the side with fewer pairs, and side was assigned from channel-name prefixes rather than coordinates.

stereo-EEG · SEEG · epilepsy · seizure-onset zone · lateralization · interictal features · implantation bias