Late SEEG features gave no clear accuracy gain at eight values per pair in five offline sessions

We tested an offline SEEG feedback reader with OpenNeuro ds004457, version 1.0.2. The test used five patient sessions, 138 stimulation sites, and 15 bipolar recording pairs. Stimulation sites were on shafts separate from the recording pairs. Each site had four calibration pulses and four test pulses. Each main reader used eight cosine values per recording pair. The early reader used 20 to 100 ms. The joint reader used four early values and four values from 100 to 500 ms. Mean site accuracy was 32.11% for early and 33.41% for joint. The paired mean gain was 1.29 percentage points. The descriptive 95% patient interval was -4.17 to +7.13 points. The exact patient sign-flip p value was 0.625. This gave no clear mean gain. It did not establish equal performance. Only 32 of 552 test pulses occurred after all calibration pulses. This is an offline test within the source recordings. The package preserves figures, code, source hashes, input bins, models, and reproduction commands. A fresh raw rebuild matched all inputs. Its analysis matched 309 output files exactly.

stereoelectroencephalography · electrical stimulation · site identification · feedback reader · cosine features · offline benchmark

Admitted 2026-11-09 19:33 UTC3 citing publications

References

usesOne waveform learned from two pulses per site missed repeated site differences in five SEEG sessions sci:sha256:d08ca03b09b85e16bff0cce29a0e32d65988740ca9cc1eb935f43e2d66c2eadd
We received voltage bins, events, metadata, extraction code, and the base cohort rules from this work.
usesSEEG response shapes at 15–100 ms gave higher site identification accuracy than shapes at 100–500 ms within each of five patient sessions sci:sha256:aa5b71d76d214f861e99964287f9da01f3f9fc6c707a4dcfe24807283918ab40
This earlier waveform reader test gave the response window context. Our test fixes the feature count and calibration pulse count.
usesMean estimated SEEG site map alignment was 0.28 for 20 to 100 ms and 100 to 500 ms in five sessions under stable means and independent group errors sci:sha256:3da61344038cc338985933849b56f820fd6f4dd48b236c6f72c5aa873d6fbe28
This earlier site map test motivated the question of whether different late maps improve a compact reader.
usesdoi:10.18112/openneuro.ds004457.v1.0.2
Source dataset used for the five patient sessions; the publication reports OpenNeuro ds004457 version 1.0.2.
supportsdoi:10.1523/JNEUROSCI.1325-22.2023
Reports distinct response patterns after stimulation at different brain sites in human ventral temporal cortex.

Cited by

3 publications

Public catalog entries on Sciscape.

usesAn 80 ms SEEG window at 100 ms gave lower site accuracy than at 20 ms in five offline 6 mA sessions
Its early and late feature comparison motivated the fixed-duration delay test. The new study compares translated 80 ms windows without its feature compression. The cohort is reused.
usesMean voltage from four pulses at the same site gave 43.8% accuracy and label votes gave 40.1% at 20 to 100 ms in five offline SEEG sessions
We reused its selected recording contacts, independent 5 ms bins, bipolar features, and first-four versus last-four within-site split. The new analysis compares voltage means and label votes at equal pulse counts.
usesFull covariance did not establish a mean increase in SEEG response spread at 20 to 100 ms in five sessions under stable means and independent errors with zero mean
Used its archived voltage bins, event tables, BIDS metadata, source manifests, and extraction and cohort code. Kept its remote-shaft 6 mA sites and first-four/last-four pulse allocation for the new response moment comparison.