In 8 SEEG patients, distance-corrected CCEP N1 latency was more consistent for stimulation pairs than recording contacts

Question: after control for distance, is the N1 latency of a cortico-cortical evoked potential (CCEP) a property of the stimulation pair (sender) or of the recording contact (receiver)? Data: expert-checked N1 peaks, contact coordinates and channel status from OpenNeuro ds005448 v1.0.0 (STReEF). The primary cohort is 8 SEEG patients. We used the input files and exclusions of AutoSci publication 0952f1d5. Method (pre-registered): for each patient, we removed log10(distance) and same-shaft effects with OLS. Then we calculated a count-matched split-half reliability (means of 3 responses, 500 repeats, Fisher z) for each receiver across disjoint halves of stimulation pairs, and for each sender across disjoint halves of recording contacts. We used exact sign-flip tests and a Holm correction. Result: receiver z = 0.20 (8/8 > 0) and sender z = 0.49 (8/8 > 0). The difference is -0.29, and the sender is larger in 7 of 8 patients (Holm p = 0.023). The pre-registered decision is 'sender-dominated'. The approximate single-response share of residual variance is 0.22 for the sender and 0.08 for the receiver. These values are stable for 2, 3 or 5 responses. The result is the same with a cubic distance model, with different-shaft responses only, in 5 ECoG patients (same direction, p = 0.063) and in all 13 patients (p = 0.0012). Shuffled residuals give z near 0. The apparent speed for different-shaft responses is about 2.3 m/s (0.44 ms/mm). Limits: 8 patients from one centre, one N1 labelling, and only responses with an N1. The test does not identify the cause of the sender effect (for example tissue, stimulus strength, artefact or labelling for each pair). Spatially smooth effects can partly enter the two reliabilities.

cortico-cortical evoked potentials · N1 latency · SEEG · epilepsy · split-half reliability · stimulation pair · recording contact