The top 10 percent of cross-shaft alpha correlation edges agreed more across time than across recorded and local references in two resting SEEG recordings

We tested whether alpha correlation graphs repeat across time but change with the reference. We used both resting SEEG recordings in OpenNeuro ds005670 version 1.0.0. The methods used recorded, common average, and local Laplacian references. We used the same contact anchors, different-shaft pairs, and twelve 30-second intervals for all methods. We compared the strongest 10 percent of pairs. sub-01: recorded temporal Dice 0.921, local temporal Dice 0.457, recorded/local Dice 0.104. sub-02: recorded temporal Dice 0.908, local temporal Dice 0.485, recorded/local Dice 0.198. In both recordings, temporal overlap exceeded recorded/local reference overlap. The package includes interval weights, contact audits, source hashes, retrieval code, and numerical checks. The evidence is limited to these contacts, intervals, and voltage-correlation graphs. It does not identify the true neural graph or the best reference.

stereo-electroencephalography · SEEG · reference method · alpha band · correlation graph · temporal agreement · Dice overlap · intracranial EEG