SEEG neighbours shared session changes no more than 30 s noise in 64 RAM FR1 patients

We tested if the change of an SEEG pair spectrum from session 0 to session 1 also occurs in the neighbour pair on the same shaft more than short-term noise does. We used archived spectra (AutoSci 72aef6af; OpenNeuro ds004789, RAM FR1; two 30 s halves per session). A plan was hashed before the analysis. For each patient, we scaled the change of 7 features over pairs and correlated it between pairs (i,i+1) and (i+2,i+3) on one shaft (S). The between-session change used different sessions; the within-session change used the two 30 s halves of one session. 64 patients (4168 couples) were included. Neighbours shared the session change (S 0.26-0.35, Holm p = 0.0007), but they shared 30 s noise by the same amount (S 0.23-0.40). S_between minus S_within was -0.05 (95% CI -0.11 to 0.03) for log total power and 0.02 (-0.06 to 0.09) for the aperiodic exponent; no feature passed (Holm p >= 0.10). A patient-wide shift was larger between sessions than within a session (6 of 7 features, not corrected). Thus we found no regional slow change beyond the spatial structure of short-term fluctuations. Limits: one task, same implant, unknown session gap, 30 s segments.

stereo-EEG · spectral change · session variability · neighbour correlation · short-term noise · RAM FR1