Single-pulse stimulation artifacts localized SEEG contacts to a median 9.7 mm and detected label swaps in eight patients

We asked whether the volume-conducted artifact of clinical single-pulse electrical stimulation (SPES) can estimate stereoelectroencephalography (SEEG) contact positions and detect label errors without imaging. We analyzed 60 pulses per patient from eight STReEF patients (717 contacts), using CT coordinates as truth. A homogeneous two-point-source model localized held-out contacts with pooled median error 9.7 mm, compared with 26.8 mm for the centroid baseline; artifact error was smaller in all eight patients (one-sided exact sign-flip p = 0.0039) and for 94% of contacts. Median error was 7.6 mm within 10 mm of a stimulated contact and 30.3 mm beyond 20 mm. The model detected 99.7–100% of tested label swaps between shafts and 93.3–100% of adjacent-contact swaps. The model explained only part of the variance (per-patient median R² 0.28–0.51), and peak amplitudes did not follow the expected 1/r falloff. The approximately 1 cm localization error exceeds the 3.5–5 mm contact pitch. These results support artifact-based checks of coarse geometry and gross label errors, but the method cannot replace CT localization. Data were from one center and used monopolar biphasic 1–2 mA clinical pulses; CT coordinates were treated as truth.

stereoelectroencephalography · SEEG · single-pulse electrical stimulation · stimulation artifact · contact localization · electrode labels · brain-computer interface · volume conduction