Two-contact SEEG phase offsets remained 63–71° after calibration in 21 patients

In a best-case, non-causal analysis of 60-s intracranial EEG segments from 21 awake epilepsy patients in OpenNeuro ds005624, we measured phase differences between adjacent-contact bipolar pairs on the same shaft. After estimating and removing a fixed circular-mean lag from the first 29 s, the median absolute phase error in the last 29 s at a two-contact separation was 63° (delta), 64° (theta), 66° (alpha), and 71° (beta). No subject had an ungated calibrated error below 45° in any band; amplitude gating reduced group errors to 48–56°. Errors increased with separation in all 21 subjects. Significant imaginary coherency occurred in 10–12% of two-step pairs, compared with 0% after the epoch-shift null, indicating consistent non-zero lag in a minority of pairs while most coupling was zero-lag or weak. These findings suggest that phase at a neighbouring pair is an unreliable proxy for phase-targeted stimulation at a pair two contacts away; reading at the target or calibrating polarity at an adjacent pair is preferable. The analysis used only 60 s per subject, included 21 of 24 patients, and used zero-phase filters, so it represents a best case without estimator delay. Contact steps are not distances in millimetres, and no stimulation was performed.

stereo-EEG · SEEG · phase error · phase-targeted stimulation · bipolar recording · imaginary coherency · intracranial EEG · closed-loop neuromodulation