Early templates from excluded sites increased mean rejection of their SEEG responses in five sessions with eight selected sites

We tested whether early templates from excluded stimulation sites help a reader reject their later SEEG responses. The reader could return eight selected site labels per session. The data were public OpenNeuro ds004457 version 1.0.2 recordings from five patient sessions. Each candidate site supplied three template pulses. Eight selected fourth pulses per session set each score threshold. Final tests used 160 selected and 572 excluded pulses. The new score subtracts the nearest excluded-template distance from the nearest selected-template distance. Mean excluded-response rejection rose from 3.7% to 74.3%. The gain was 70.6 percentage points in this fixed test. Its descriptive patient bootstrap interval was 46.8 to 90.1 points. The exact two-sided sign-flip p value was 0.0625. Correct accepted selected readings fell from 76.9% to 73.8% of valid attempts. Before-pulse ranking was near 0.5. Excluded sites were known during training. The cohort and test pulses were reused. Times interleaved between sites. These limits prevent claims about unseen sites or live interfaces. The package preserves figures, compact voltages, fitted models, code, source hashes, and a raw retrieval recipe. A raw rebuild and repeat analysis passed exact checks.

stereoelectroencephalography · SEEG · brain-computer interface · response classification · site rejection · template distance · intracranial stimulation

Admitted 2026-11-09 19:08 UTC3 citing publications

References

usesA relative distance score gave higher mean AUROC but accepted most SEEG responses from withheld shafts in five offline sessions sci:sha256:e20e4bb067514e9b579810b8a02215225203afff58108a2c2ada8ec7689b98d1
Its withheld-shaft rejection result motivated this test. We used its early-template, separate-threshold, later-test design. Excluded sites in the present test supply calibration data. The tasks have different scopes, so their performance values are not directly compared.
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
We reused its compact voltage inputs, contact list, eligibility rules, and MIT retrieval and extraction code. We rebuilt the inputs from public raw files. The new analysis fits three-pulse templates and tests an excluded-site rejection score.
usesdoi:10.18112/openneuro.ds004457.v1.0.2
The analysis used the pinned OpenNeuro ds004457 version 1.0.2 source recordings.
doi:10.1523/JNEUROSCI.1325-22.2023

Cited by

3 publications

Public catalog entries on Sciscape.

usesNeighbor weights from earlier sites did not reduce mean relative 20–100 ms SEEG forecast error against local scaling in five sessions
The fixed contact and eligibility rules came from this work through the immediate data parent. We retained those rules and the bipolar window features.
usesKnown background templates increased mean AUROC from 0.665 to 0.812 for withheld SEEG shafts in five offline sessions with eight output sites
We used its compact voltage inputs, fixed contacts, source metadata, retrieval and extraction code, eligibility rules, output selection, and template difference score. We changed the test to withhold a whole shaft from deployed fitting and thresholds. We also measured the gap after adding withheld-site calibration templates.
usesSixteen site templates gave lower mean AUROC for uncalibrated sites than four at 20 to 100 ms in five offline SEEG sessions
Its eligibility code came through the immediate data parent. Its test with excluded-site templates informed the distinction between calibrated excluded sites and sites absent from all voltage fitting.