Five templates per site certified 66.46% of SEEG search labels in eight patients on a ±9.77 ms grid under exact arithmetic
We tested a compact reader on public SEEG data from eight STReEF patients. Each patient supplied four stimulation sites and five later test pulses per site. A full reader searched 41 fit templates per site. Five retained templates and line error radii certified 4,360 of 6,560 query views (66.46%) across a ±9.77 ms grid. Every certified label matched the full reader. The descriptive patient bootstrap interval was 52.84% to 80.73%. The proof requires a fixed bank, fixed scales, and exact arithmetic. Separate long-double checks matched the float64 results. An implemented fallback kept all full-reader labels with 58.36% fewer length-32 query-template dot products. Compact coordinates plus radii fell by 87.50%; a fallback still needs the full bank. Of the 4,360 certified labels, 857 were wrong for the recorded site. Control-window labels also received certificates. Thus agreement does not prove a true site or neural signal. The package keeps native crops, source pins and hashes, code, proof, figures, bounds, and replay instructions.
stereoelectroencephalography · SEEG · signal processing · template reader · distance bounds · timing error · epilepsy
References
5- usesSelection for distinct response patterns had an uncertain mean accuracy gain at four SEEG pairs in eight STReEF patients sci:sha256:143d7f91f05a486692072b6c1e61b67efa46729f12681baae0430cec9016d9ad
Used the metadata selection and pinned raw-range retrieval helpers through the immediate input publication. The fixed four-site sample and early/later split originate in this work. This is the same cohort, not an independent replication.
- usesA template search improved SEEG condition-label robustness across an imposed ±9.77 ms error grid in eight patients sci:sha256:e706b9b7d629116692d403dfd19f62041a02b8c4b778b10358cc9e37db0c249d
Used the exact native voltage crops, fixed eight-patient sample, source metadata and pins, and feature and full-reader code. The full baseline was replayed exactly. The new method bounds that fixed reader with retained timing anchors and line error radii.
- usesdoi:10.18112/openneuro.ds005448.v1.0.0
The analyzed SEEG recordings and metadata are from STReEF, OpenNeuro ds005448 version 1.0.0.
- doi:10.1016/j.ynirp.2025.100274
- doi:10.1007/s12021-022-09567-6
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesAn offline Fourier bound certified a 34.6% mean SEEG match rate in eight patients for offsets up to 10 ms under exact arithmetic
Its finite template-search certificate frames the earlier timing question for a response reader. The present method tests a fixed template threshold over a continuous interval in a different source cohort. The acceptance rates have different meanings.
- usesFive time templates reduced SEEG site log loss with simulated grid errors up to 10 ms in five patients
This related search study informed the use of a finite bank of pulse-time templates. Our test uses ds004457 and compares site probability rules under artificial grid errors, rather than certification of search labels.