In 23 RAM FR1 patients, bipolar SEEG spectral patterns were stable across sessions but less reliable than within sessions

We compared 60-s bipolar SEEG spectral features across two FR1 sessions in 23 patients from OpenNeuro ds004789 (2,399 contact pairs). Six features reproduced across sessions: mean Spearman rho ranged from 0.725 for relative theta power to 0.923 for 70–150 Hz log power. For every feature, across-session reproducibility was lower than the Spearman–Brown corrected 60-s within-session value (22 or 23 of 23 patients; Holm-adjusted p = 6e-5 for each). Pair fingerprinting identified the matching contact pair in 36% of across-session comparisons (95% CI 30–42%; chance 1.1%). These results indicate that session-to-session variance limits the reliability of one-segment normative maps and contact biomarkers; theta was least stable. The two within-session halves were adjacent in time, the interval between sessions is unknown, task phase was uncontrolled, and seizure-onset contacts and interictal spikes were retained. One subject with byte-identical session segments was excluded after a partial run; other partial duplication was not assessed. The analysis included the first 24 eligible subjects in sorted order, with 23 retained, from one RAM centre group.

SEEG · spectral features · test-retest reliability · intracranial EEG · FR1 · OpenNeuro · bipolar recordings