Thalamic recruitment followed the seizure-onset zone by a mean 23.5 s; timing and recruited fraction did not differ from white matter in 53 SEEG seizures

We measured when ipsilateral thalamic SEEG pairs start a sustained increase of 1-80 Hz line length after the marked seizure onset. We used OpenNeuro ds007445 v1.0.4 (19 patients, first 3 seizure files per patient, 53 seizures, 3550 labelled bipolar pairs). The plan was written and hashed before we read the signals. For each pair, the latency is the first 1-s bin in [-10, 50) s with robust z > 5 for 3 bins, against the baseline [-120, -70) s. The region latency is the median over pairs, with censoring at 50 s. The patient is the unit, and the tests are exact sign-flip tests with Holm correction. The thalamus started later than the seizure-onset zone in 17 of 17 patients (mean +23.5 s, Holm p = 2.3e-5). Thalamus minus ipsilateral white matter latency was +6.9 s (12 of 15 non-zero patients later, Holm p = 0.27). The thalamic recruited fraction was 0.13 lower than white matter (Holm p = 0.27). The thalamic median latency was below 50 s in 20 of 51 seizures and below 10 s in 13 of 51. The same detector before onset marked 9% of thalamic pairs. AN and CM latencies were the same (p = 0.98). Thus, with this detector, the data do not show early or thalamus-specific recruitment. Thalamic involvement agrees with general propagation. The package has the code (numpy only), byte-range provenance with SHA-256, per-pair z time courses, tables and instructions.

stereo-EEG · seizure onset · thalamus · recruitment latency · line length · epilepsy