Seizure termination channel order was reproducible across seizures in 30 HUP SEEG patients; onset order did not fully explain it

Question: Is the channel order of seizure termination reproducible across seizures of the same SEEG patient, and is it as reproducible as the onset order? We analyzed archived line-length features for 132 ictal SEEG runs from the HUP iEEG dataset (OpenNeuro ds004100, version 1.1.3). Channel onset and termination times were identified with the line-length detector from prior work (robust z score, primary threshold K=5). For every pair of usable seizures in a patient, we calculated Spearman correlations across common active channels using average ranks for ties. The patient was the unit of analysis; primary tests used sign flips and Holm correction. In 30 patients, 115 seizures and 175 seizure pairs, mean termination-order correlation was 0.361 (95% CI 0.280 to 0.437); 27 of 30 patients had positive values (p<0.00001). Mean onset-order correlation was 0.481 (95% CI 0.379 to 0.577). The termination-minus-onset difference was -0.122 (95% CI -0.249 to 0.011; Holm p=0.086), so onset order was not significantly more reproducible in the primary analysis. After removing onset rank, termination order remained correlated across seizures (mean 0.309; 27 of 30 patients positive). A post hoc control for channel involvement strength gave 0.248 (26 of 30 positive). Termination order was reproducible in sensitivity analyses at K=3 and K=8, among uncensored channels, and using the first two seizures. The difference between onset and termination reproducibility was significant only at K=8 (uncorrected p=0.018). This single-centre study used one line-length detector with 0.5-s resolution; stable channel properties may partly explain termination order, and secondary and post hoc tests were not corrected for multiple comparisons.

stereo-EEG · seizure termination · channel order · reproducibility · line length · epilepsy · Spearman correlation