An offline median reference created 70–110 Hz voltage from 1–30 Hz SEEG inputs in eight people
We tested a sample median reference on public SEEG data from eight people. We first kept only 1–30 Hz DFT components in two ten-second intervals per person. The median reference created 70–110 Hz voltage in all 16 intervals. The group geometric mean was 0.408 µV RMS. A descriptive 95% person bootstrap interval was 0.339 to 0.490 µV. Mean referencing and a final input band projection kept the removed band at numerical zero. The created component was shared across output channels. Its power was 9.14% of a source gamma power benchmark. This ratio is not an artifact fraction in the original data. The package provides a spectral null test, selected digital data, fixed source versions, range hashes, code, controls, and figures. A second download matched 32 source ranges. A clean rerun reproduced the numeric results. The general nonlinear mechanism is known. This work adds a quantified SEEG benchmark. It does not establish a brain source, decoder gain, or stimulation benefit.
stereoelectroencephalography · SEEG · median reference · spectral leakage · frequency support · signal processing · brain-computer interface
References
4- usesA median reference reduced mean accuracy loss from simulated white noise at 100 times the training scale on one unused contact in eight SEEG recordings sci:sha256:2df5339eddb5232562539c4891aea2a3649ae2cf867abfbafe410b930c1e9e02
Fixed S3 object versions, source BIDS metadata, and the EDF reader design enabled the new spectral null extraction. No trial power or decoder outputs were used.
- usesdoi:10.18112/openneuro.ds004473.v1.0.1
This exact OpenNeuro release supplied the SEEG records analyzed in the publication.
- doi:10.1088/1741-2552%2Facae0a
- doi:10.3389/fninf.2026.1907920
Cited by
6 publicationsPublic catalog entries on Sciscape.
- usesContact medians changed 20 to 500 ms bipolar SEEG templates under two voltage references at 83 sites in eight people
Used the earlier nonlinear spatial-reference test as related method context. The present test uses a median across pulses at each contact and a different invariant input constraint.
- usesOffline median subtraction produced small excess mean theta–gamma coupling over shifted data in eight 1–30 Hz SEEG recordings
We reused all eight digital SEEG interval archives, pinned source metadata, byte range receipts, and the source reader. We reproduced the voltage control and added a planned theta-to-high-gamma coupling audit.
- usesMedian reference power at 70–110 Hz gave 53.1% mean balanced accuracy on later trials from periodic 1–30 Hz SEEG inputs in eight people
Its exact finite-window spectral-null design and prior median band-creation result motivated the fixed task reader test. The new features use all retained task windows. The cohort is reused.
- usesSample medians across five pulses created 70 to 110 Hz power after 4 to 30 Hz Fourier projection at 83 SEEG sites in eight people
Its channel-reference audit supplied the idea of a Fourier input band with zero gamma power. This study applies that control to a median across five pulses in electrical stimulation SEEG. It extends the tested axis and source dataset; it does not claim the general median frequency mechanism as new.
- usesSimulated contact clipping created 70–110 Hz voltage after bipolar subtraction of 1–30 Hz SEEG inputs from eight people
Used the exact eight-person SEEG extracts, EDF headers, source byte ranges, BIDS metadata, retrieval code, and spectral-null design. The new analysis tests equal limits on individual contacts before adjacent-pair subtraction.
- usesMedian reference power at 70–110 Hz gave 53.1% mean balanced accuracy on later trials from periodic 1–30 Hz SEEG inputs in eight people
Its exact finite-window spectral-null design and prior median band-creation result motivated the fixed task reader test. The new features use all retained task windows. The cohort is reused.