Shared contacts accounted for a mean 49.2 percent of maximum absolute correlation tree edges in 1–80 Hz bipolar SEEG from 36 patients

We tested correlation trees in public SEEG from 36 patients in OpenNeuro ds004100 version 1.1.3. Each patient supplied two interictal runs with two 30-second blocks per run. We formed consecutive bipolar pairs and maximum trees from absolute Pearson correlations. In 1–80 Hz data, links that shared an input contact accounted for a mean 49.2 percent of tree edges. The descriptive patient bootstrap interval was 46.8 to 51.5 percent. Shared contacts accounted for 2.3 percent of all possible links. Mean run retention was 89.5 percent for shared-contact tree edges and 57.2 percent for disjoint-contact edges. An exact control kept contact variances and set cross-contact covariance to zero. We proved that every maximum tree retains all positive edges of the contact overlap forest and adds zero-weight bridges. The control tree share was 85.6 percent. Tied bridge choices changed tree shape in all patients. The measured counts do not identify the neural or nonneural cause of each edge. The package preserves input covariance, source pins, code, all tree choices, figures, and reproduction instructions. This study does not test clinical benefit.

stereo-EEG · bipolar signals · spanning trees · contact overlap · Pearson correlation · network analysis