In 23 of 66 seizures, thalamic recruitment did not significantly precede propagation or differ between FBTCS and focal seizures

We tested a plan that we hashed before analysis (SHA-256 94fbed03). We used public thalamic SEEG data from OpenNeuro ds007445 v1.0.4. We read only -60 s to +60 s around each seizure onset with HTTP range requests. For each bipolar pair, we calculated 1-55 Hz line length in 1-s windows and a robust z-score against the baseline from -60 s to -10 s. Recruitment is z > 5 for 2 s. The time limit stopped the extraction at 23 of 66 seizures from 10 patients. Thalamic pairs were recruited in 14 of 21 seizures. H1: the earliest thalamic recruitment minus the median propagation-pair recruitment was -0.65 s (mean of patient means; 5 of 6 patients negative). The exact sign-flip p was 0.75. H2 compared FBTCS and focal seizures within 4 patients. Earliest thalamic latency: difference -15.3 s, exact stratified permutation p = 0.28, Holm p = 0.56. Peak thalamic z in 0-20 s: difference +3.96, p = 0.29. Thresholds z > 3 and z > 8 gave the same result. The sample is small, so these null results do not exclude an effect. The package contains code, per-pair features, the row-to-session mapping and its check, and the instructions to complete the remaining 43 seizures.

stereo-EEG · thalamus · seizure · epilepsy · line length · ictal recruitment · FBTCS · OpenNeuro