Separate pair normalization kept more SEEG site decisions correct under independent constant positive output gains within ±6 dB in five sessions
We tested whether separate normalization of each bipolar recording pair keeps SEEG site decisions correct under independent output gains. We used public OpenNeuro ds004457 version 1.0.2 data from five sessions. The test included 183 sites and 732 test pulses. Each gain was positive and constant within one response window. Calibration stayed fixed. At every gain vector within the ±6 dB box, 35.8% of pulses stayed correct with separate pair normalization. The matched common normalization reader kept 14.6% correct. These are equal-patient means. The paired gain was 21.2 percentage points. Its descriptive bootstrap interval was 10.3 to 29.3 points. The exact two-sided patient sign-flip value was 0.125. With no gain change, accuracy fell from 38.3% to 35.8%. The proof shows why separate pair scores remove positive gains. An exact affine box test finds common-reader failures. The benefit remained with a later window, no shared recording contacts, and fewer calibration pulses. Fresh extraction from 110 hash-checked public files matched all input arrays. A repeat run matched all 25 models and the result tables. The package preserves data, code, fitted models, four figures, the proof, and reproduction instructions. The five sessions were used in prior work. This offline stress test does not establish a population effect. It does not test new noise, contact gains, physiological change, perception, or live control.
stereoelectroencephalography · electrical stimulation · bipolar recordings · gain invariance · cosine normalization · signal processing · patient bootstrap
References
4- usesCovariance angles kept 37.9 percent mean site accuracy under nine simulated positive common voltage gains in five offline SEEG sessions sci:sha256:0dbd4e1ba615799ff752cce0c46663f54b7af5ebdf8dd5a55d3ee4cebeb486e4
This publication supplied the common covariance cosine reference and the reused pulse split. All no-gain reference predictions matched. The new work tests separate pair normalization and correctness throughout an independent gain box.
- usesNoise correction did not increase mean eight-site SEEG accuracy at 20–100 ms with four calibration pulses per site in five sessions sci:sha256:eef87e94853fad626caddee9fff3146fa92c84241f432ff4a03376aaf5f27399
This publication supplied the public voltage bins, event tables, source manifest, and selection code. Fresh extraction checked all input arrays before the new pair gain analysis.
- usesdoi:10.18112/openneuro.ds004457.v1.0.2
This is the exact OpenNeuro snapshot version used for the archived recordings and source metadata.
- doi:10.1523/JNEUROSCI.1325-22.2023
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesAbsolute pair scores kept SEEG site decisions fixed under simulated output sign reversals in five sessions
This work supplied the fixed public voltage bins, event tables, source hashes, input-selection code, and separate-pair covariance cosine rule. We add absolute scores, output sign stress tests, and an unlabeled fitted-reader tool.
- usesContact normalization kept SEEG site decisions fixed under simulated five percent contact gain changes but reduced accuracy in five offline sessions
Supplied the separate-pair gain normalization rule and the per-pair covariance rule through the input code. Its gain acted after bipolar subtraction. The present work tests contact gains before subtraction.
- usesSimulated joint contact gain errors within five percent changed 172 SEEG decisions that passed separate contact tests in five offline sessions
The independent bipolar output gain audit supplied a related affine box bound. We apply that standard bound to pulled-back contact score terms before bipolar subtraction and compare it with separate contact tests. We do not claim the general box formula as new.