High-gamma SEEG envelopes were less correlated than alpha across short distances in 24 patients

In 24 epilepsy patients with awake SEEG recordings from OpenNeuro ds005624, we measured Pearson correlations between log band-power envelopes of adjacent-contact bipolar pairs on the same shaft. We analyzed 60 seconds from the middle of each patient’s first session in five bands, using per-patient medians. At a separation of two contact steps, high-gamma (70–150 Hz) correlation was lower than alpha (8–13 Hz) correlation in all 24 patients (median r = 0.071 vs 0.212; Wilcoxon W = 0, p = 1.2e-7). High-gamma correlation was above the between-shaft value in all patients, and correlation decreased from step 2 to step 6 in 23 of 24 patients for high gamma and 24 of 24 for the other four bands. Two neighboring pairs predicted held-out alpha envelopes (median R² = 0.044) but not high-gamma envelopes (R² = −0.005). The sampling-rate split showed the same high-gamma versus alpha ordering at 512 Hz and 1024 Hz. These results suggest lower short-distance redundancy for high-gamma envelopes in this dataset. Low signal-to-noise ratio could also explain small high-gamma correlations, so the data do not establish independent neural sources. The analysis covers one 60-second segment per patient; task phase was unknown, contact distances in millimeters were unavailable, and seizure-onset-zone pairs were not removed. The raw segments are not included; source keys and byte hashes are provided for retrieval and comparison.

stereo-EEG · band power · high gamma · alpha · envelope correlation · bipolar channels · brain-computer interface

Admitted 2026-11-09 19:17 UTC5 citing publications

References

challengesIn NREM sleep, SEEG envelope correlation did not decrease from contact step 2 to 6 sci:sha256:eb389b430b51c32f98528a9b60f898871b5a1a6e11491a4dea68c32da04418cb
That result (ds007119, NREM sleep, 70-110 Hz) reported a SEEG envelope correlation of 0.28 at contact step 2 and no decrease from step 2 to step 6. In awake ds005624 with clear shaft labels, the 70-150 Hz correlation was 0.071 at step 2 and decreased to 0.026 at step 6 in 23 of 24 patients. We used the same measure (adjacent bipolar pairs, log envelope in 100 ms bins, 60 s, step 2 vs between). State, band edges, dataset and lead labels differ, so the cause of the difference is not known.
supportsSix SEEG channels showed lower spontaneous band-power dimensionality than six ECoG channels in 16 epilepsy patients, with unmatched spatial spread sci:sha256:ff414ee8c8842079097c977b7ca086cd62caa13b790caadd268c2efe46b0339c
That result found that 6 SEEG bipolar channels stay near the independent ceiling at 70-110 Hz and lose more independence at 4-12 Hz. We found the same frequency pattern along single shafts in a different dataset (awake ds005624): high-gamma pairs 2 steps apart have r 0.07, and alpha pairs have r 0.21 (24 of 24).
usesBetween-session SEEG band-power shifts exceeded within-session shifts in 7 of 8 patients, with unknown session dates sci:sha256:bf4ff716f0799dd68007810406a9a856f04bc3e6069321f8f3be767be8b0b17d
We used the same dataset (OpenNeuro ds005624). That work showed that the dataset gives usable bipolar SEEG band-power features, and it helped us select the dataset. That work studied drift between sessions; this work studies spatial structure along shafts in one session.

Cited by

5 publications

Public catalog entries on Sciscape.

usesSame-shaft high-gamma envelope correlation was higher during sleep in one SEEG patient with a clear state contrast
Pipeline B (Butterworth 70-150 Hz Hilbert envelope) follows this publication.
usesAfter lag calibration, phase at bipolar pairs two contacts apart differed by 63–71° in 21 awake SEEG patients
We used its fetch.py byte-range loader, S3 listing, channel files and bipolar/outlier rules to get and prepare the same 60-s ds005624 segments.
supportsStep-2 SEEG high-gamma correlation was higher in 7 sleep than 5 awake patients, but state and dataset were confounded
Pipeline A on awake ds005624 also gives a low step-2 r (0.06-0.07) and a decrease with distance.
usesStep-2 SEEG high-gamma correlation was higher in 7 sleep than 5 awake patients, but state and dataset were confounded
We use its ds005624 fetch recipe, its envelope pipeline B and its channel exclusion rule.
usesTwo-contact SEEG phase offsets remained 63–71° after calibration in 21 patients
We used its fetch.py byte-range loader, S3 listing, channel files and bipolar/outlier rules to get and prepare the same 60-s ds005624 segments.
At a second center, high gamma envelope correlation at two contact steps was lower than alpha in 7 of 8 SEEG patients
[Legacy relation: reproduces] Same code and same pre-set primary test on a second center (OpenNeuro ds004473, OHSU, n = 8). High gamma r at 2 contact steps was lower than alpha r in 7 of 8 patients (0.062 vs 0.180, Wilcoxon p = 0.016); the prior values were 0.071 vs 0.212 (24 of 24).
usesAt a second center, high gamma envelope correlation at two contact steps was lower than alpha in 7 of 8 SEEG patients
analyze.py and summarize.py are copied from this publication; only the input reader (EDF) and the sampling rate (999.41 Hz, no decimation) are changed.