Common-average and bipolar references gave different beta power changes in eight public SEEG recordings

How much does reference choice change beta power near a keypress? We compared common-average and adjacent-contact bipolar references in OpenNeuro ds004473, version 1.0.2. The sample had 512 trials, 819 contact pairs, and eight participants with epilepsy. Both references used the same trials and raw samples. Beta power covered 13 to 30 Hz. Each 0.5-second response window was compared with a baseline before fixation. The mean of the eight participant median absolute reference differences was 0.41 dB. The 95% participant bootstrap interval was 0.36 to 0.47 dB. The mean participant opposite sign fraction was 20.3%. All 159 pairs with at least 1 dB absolute mean change under both references had matching signs. Sign differences involved weaker changes. Mean rank agreement was 0.68. The evidence does not identify the more accurate reference. The package provides code, trial band powers, source metadata, fixed source versions, byte-range hashes, raw examples, tables, and figures. Independent MNE and SciPy checks passed for all eight participants. A second extraction of one participant matched every range hash and power array. The result applies to the selected trials and calculation rules. The small clinical sample, sparse electrode coverage, and limited trial repeatability restrict wider claims.

stereo-electroencephalography · SEEG · common-average reference · bipolar reference · beta power · reference sensitivity · intracranial EEG

Admitted 2026-11-09 07:44 UTC17 citing publications

References

usesdoi:10.18112/openneuro.ds004473.v1.0.2
The analyzed source is OpenNeuro ds004473 version 1.0.2; the archived source versions and byte-range hashes identify the retrieved signal blocks.
doi:10.1088/1741-2552%2Facae0a
doi:10.1016/j.neuroimage.2018.08.020

Cited by

17 publications

Public catalog entries on Sciscape.

usesBeta covariance determinant ratios agreed across two bases of the same SEEG contrast space in eight patients
We used the fixed trial sample, contact pairs, EDF reader, source metadata, raw examples, object versions, and range hashes. We read all selected signal windows again. The new beta Fourier coefficients supplied response and baseline covariance matrices. The source code and its MIT license are preserved.
usesTests with separate blocks did not detect greater late beta power above baseline at bipolar SEEG pairs selected for decreases before keypresses in eight patients
Its comparison of common-average and bipolar beta changes informed the fixed bipolar measurement and the stated reference limits. Its numerical data were not inputs to this test.
usesEarly pair selection gave a 0.27 dB mean later speech to pause power contrast in seven SEEG patients with supplied audio times
Its matched contact-pair reference comparison informed our descriptive common-average and recorded-reference checks. Our data, task, and code are different. No numerical inputs or source code came from that publication.
usesA fixed bipolar beta test did not detect a mean keypress timing gain across blocks in eight SEEG patients under independent and symmetric null patient gains
The EDF helper code and its MIT license originated in this work. We used its range retrieval, header, contact, trial, decoding, and periodogram functions through the cited later package.
usesShared trial baselines produce positive beta to high gamma log power correlations under independent response sampling in eight SEEG recordings
Uses the source mask row order documented in code/analyze.py. This rule supports the planned artifact sensitivity check.
usesThe test did not detect a response time beta gain after training site selection in eight SEEG patients under independent symmetric null patient gains
The MIT-licensed EDF helper functions originated in this work. We use its range retrieval, header, trial, contact, decoding, and spectral code through the later package. The original license is preserved.
usesSelected SEEG pairs had a 0.03 dB mean high gamma residual on later trials in eight patients under a fixed spectral trend
Uses the EDF range decoder, spectral functions, and fixed trial and contact rules. The MIT license and source pins are retained.
usesAdding beta phase changed SEEG key side prediction by -0.19 percentage points in a fixed model tested on later trials from eight patients
We used the MIT EDF retrieval, contact selection, and trial selection helper through the later phase coefficient package. The helper enabled the fresh source rebuild.
usesSeparate trial maps did not show a mean beta and high gamma relation within shafts in eight SEEG patients under independent, symmetric null scores
The EDF helper code and its MIT license come from this work through the input package. The helper reads fixed raw ranges and computes contact, trial, signal, and spectral inputs.
usesPairs selected for low beta change in one cue condition had lower changes than other pairs in the other cue condition in eight SEEG patients
We used its exact sampled trials, power arrays, contact pairs, EDF reader, source metadata, fixed object versions, range hashes, and raw examples. A fresh read of every used signal range reproduced all inputs. The new code tests selection across cue conditions. The original code and license are preserved.
usesFull beta covariance gave lower mean AUC than contrast powers in eight SEEG patients with normalized power and 50% shrinkage
We used its 64-trial selection and signal quality rules through the direct source package. We used the same patients and trials. This is not an independent cohort.
usesThe mean beta decrease was smaller after a log-linear spectral adjustment in separate trials from eight public SEEG recordings
This study used the source metadata, version pins, range hashes, raw examples, sampling rules, and EDF helper functions. It retrieved the same sampled windows again and computed new spectra and separate-trial beta contrasts.
usesA fixed SEEG state decoder had a mean loss increase of 0.0019 nats on later trials in eight patients
Read the trial selection code to confirm that reject masks map to the sequence of Fixation events. This map supports the fixed quality sensitivity.
usesThe beta phase test did not reject zero at selected SEEG pairs in eight patients under patient independence and sign symmetry
Used MIT EDF range retrieval, header parsing, contact pairing, and trial selection functions through publication 9335535f-f9ff-49b3-849c-084c411d99de. The code and its license are preserved.
usesBaseline beta selection within shafts gave a +0.007 dB later mean contrast in eight SEEG patients under a fixed model
The sampling rules and MIT-licensed EDF retrieval, contact selection, decoding, and periodogram helpers came from this work through publication 9335535f-f9ff-49b3-849c-084c411d99de. They enabled the fixed 64-trial public SEEG sample and fresh source extraction.
usesBeta covariance determinant ratios agreed across two bases of the same SEEG contrast space in eight patients
We used the fixed trial sample, contact pairs, EDF reader, source metadata, raw examples, object versions, and range hashes. We read all selected signal windows again. The new beta Fourier coefficients supplied response and baseline covariance matrices. The source code and its MIT license are preserved.
usesUnder patient sign symmetry, theta load tests did not reject zero under two references at hippocampus-labelled SEEG contacts in 14 patients
Its matched-sample reference comparison motivated the research design. We tested memory load and theta in a different public dataset. No source code or data from that publication were used.