Contact depth separated SOZ pairs, while high-frequency activity retained within-shaft signal after adjustment in two cohorts

A pre-registered replication tested whether interictal high-frequency activity (HFA) separates seizure-onset-zone (SOZ) from non-SOZ bipolar pairs on the same stereoelectroencephalography shaft after adjustment for contact depth. The analysis used archived per-pair feature tables from 44 patients in OpenNeuro ds003876 (27 patients) and ds005398 (17 patients), with 100 mixed shafts. Contact 1 was assumed to be the tip, and depth adjustment used a linear fit within each shaft. The primary depth-adjusted HFA AUC was 0.629 (95% CI 0.568–0.686; 30/44 patients above 0.5; Holm-adjusted p=0.0002). Depth alone had an AUC of 0.734 (0.628–0.833; p=0.0002), and raw HFA had an AUC of 0.700 (0.620–0.776; p=0.00015). Adjusted HFA AUCs were 0.588 in ds003876 (95% CI 0.515–0.662; p=0.016) and 0.693 in ds005398 (0.601–0.779; p=0.0005). SOZ pairs were deepest in 62/100 shafts. In ds003876, secondary line-length and spike-rate AUCs fell from 0.654 to 0.526 (p=0.25) and 0.640 to 0.539 (p=0.17), respectively, after adjustment. Depth explained part of the within-shaft HFA signal, while an adjusted signal remained in both cohorts. Clinical SOZ labels may follow target depth because clinicians target deep mesial structures, so adjustment may remove true SOZ signal and the adjusted AUC is a conservative estimate. Contact order was assumed, only a linear depth model was tested, tissue labels were unavailable, HFA bands differed (80–150 versus 80–250 Hz), and clips were short.

stereoelectroencephalography · seizure onset zone · high-frequency activity · contact depth · AUC · epilepsy · replication