Far-channel SEEG voltage identifies the stimulated site above chance 200–500 ms after one pulse in seven patients
In seven patients with stereo-electroencephalography (SEEG), a 12-way nearest-centroid classifier identified which of 12 bipolar stimulation sites received a single 1–2 mA pulse from far-channel voltage (channels at least 15 mm from stimulated contacts). Mean accuracy was 0.200 at 200–500 ms, above the pre-pulse mean of 0.080; all seven patients had higher late than pre-pulse accuracy (difference +0.120, exact one-sided sign-flip p=0.0078). Late accuracy exceeded the 95th percentile of 50 label shuffles in six of seven patients. Mean accuracy was 0.399 at 20–150 ms and 0.121 at 500–950 ms. Adding late features to early features did not improve accuracy (mean change −0.062; two-sided p=0.094). The apparent time course to about 650 ms was descriptive: no sliding window survived Holm correction. The analysis used seven patients, one linear classifier, and far channels only; volume conduction from near channels is not excluded. The source files were read from unversioned OpenNeuro S3 URLs, so the exact dataset release is uncertain.
stereo-electroencephalography · single-pulse stimulation · site identification · evoked response · classification · epilepsy
References
1- usesSingle-pulse SEEG responses identify stimulation sites at least 20 mm apart from distant contacts sci:sha256:2d62a5de2674244c0c30cb21e596c85383be7629c42a2746e4f8c83288088f19
Same patients, pairs, pulse rule, channel rule, preprocessing and classifier; this work extends the time window to 1 s and reproduces its 20-150 ms accuracy (0.399 vs 0.425).
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.