SEEG bolt axes predict deepest contacts within 1–2 mm for most shafts, but 13% exceed 3 mm

We measured agreement between the skull-bolt axis and the intracranial contact line for 616 stereo-EEG (SEEG) shafts in 60 subjects from OpenNeuro ds005083. The subject was the unit of inference. Median subject-level deflection was 1.72° (95% CI 1.42–1.92), entry offset was 0.29 mm, and tip error was 1.13 mm (95% CI 0.98–1.30). Tip error is the distance from the deepest contact to the extended bolt line. The distribution had a heavy tail: 9.9% of shafts had more than 5° deflection and 13.0% had more than 3 mm tip error; the 95th percentile tip error was 6.5 mm. Within subjects, tip error increased with contact span (median ρ = 0.11; 39 of 60 subjects positive; sign-flip p = 0.0074). Deflection decreased with span (median ρ = −0.19; p = 0.0002), opposite to the preregistered direction. Axis-estimation noise on short spans may explain this relation; the analysis did not separate this noise from true bending. The result does not show that long electrodes bend less. Bolt-only estimates are within about 1–2 mm for most shafts, but are not safe for all shafts. Coordinates and bolt points were taken as given, with mm RAS assumed from the Freeview files; the derivative coordinate space and bolt-point construction are not documented. The study used only trajectory derivatives and had no clinical covariates.

stereo-EEG · SEEG · electrode trajectory · bolt axis · contact geometry · tip error · OpenNeuro · ds005083