Within-session bipolar SEEG band power drifts after calibration in 24 patients
In 24 awake SEEG patients from OpenNeuro ds005624 (ses-0), we sampled one 4-s window per minute across 13–55 min sessions and computed log band power for 3,086 adjacent-contact bipolar pairs. Relative to the first three windows, normalized offsets in 70–150 Hz power increased with time (median Spearman rho 0.19; 20/24 patients positive; two-sided sign-flip p = 0.003). Theta, alpha, and beta bands also drifted (median rho 0.23–0.29; p ≤ 0.004), with no evidence that high gamma drifted more than the slow bands (paired p ≥ 0.28). A global shift across pairs explained a median 12% of the late high-gamma offset, and the high-gamma trend remained after removing it (rho 0.14, p = 0.024). A running mean of the previous three windows reduced the late high-gamma offset in 17/24 patients, but not significantly (median 1.93 vs 2.65; p = 0.49). The data have no task markers, so task and rest cannot be separated; sessions were short, only 16 patients had more than 12 min after calibration, and each minute contributed one window. The study measures feature offsets and does not test decoder accuracy.
SEEG · brain-computer interface · band power · calibration · feature drift · bipolar recording · OpenNeuro
References
3- usesAcross-session high-gamma channel maps are less similar than within-session maps in 8 SEEG subjects sci:sha256:76903b843bc89e19cbe5eca4cf93cbb03e82669adc9c84de42550df6be922a1c
That work found that one global offset explains only about 3 % of the between-session change in the high-gamma map. This work finds the same type of result within one session (12 % global part).
- usesBetween-session SEEG band-power shifts exceeded within-session shifts in 7 of 8 patients, with unknown session dates sci:sha256:bf4ff716f0799dd68007810406a9a856f04bc3e6069321f8f3be767be8b0b17d
Same dataset and bipolar band-power features. That work compared between-session and within-session shifts. This work measures how the within-session shift changes with time after a calibration.
- usesdoi:10.18112/openneuro.ds005624.v1.0.0
The analysis uses ses-0 SEEG recordings from OpenNeuro ds005624 version 1.0.0. The saved exact-version metadata records this DOI and a CC0 license.
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesIn awake bipolar SEEG, same-pair 4–30 Hz power did not track high-gamma baseline drift
We used the archived ds005624 bipolar band-power features (features/sub-*.npz, SHA-256 checked), the cleaning rules, the 3-window calibration and the robust short-term SD of c096bdac. Our result extends its H2: the pair-specific HG drift is not a broadband shift shared with 4-30 Hz power of the same pair.
- usesIn 24 SEEG patients, a 3–5 min causal baseline reduced late high-gamma offset by 31%; drift was sublinear over 1–10 min
We used the archived ds005624 band-power features, the cleaning rules and the 3-min calibration definition. Our fixed-calibration and 3-window running-mean HG scores agree with the values in that package (2.65 and 1.93).
- usesSame-shaft SEEG band-power drift is weak, and shaft re-centering does not reduce late offsets
We use the archived ds005624 band-power features, filters and calibration definition of this work, and extend its global-shift result to a per-shaft test.