Removing one shaft from the average reference replaced a mean 11.3% of top 10% beta debiased squared wPLI links and 8.7% of coherence links in 60 second samples from 34 SEEG patients

An average reference changes when channels are excluded. We tested the effect on top beta phase lag links. We used public OpenNeuro ds004100 version 1.1.3. The test used two 60 second samples per patient and 34 patients. We removed each shaft from the reference. We kept the recorded target signals and surviving pairs fixed. Mean top 10% link replacement was 11.33% for signed debiased squared weighted phase lag index. It was 8.67% for magnitude squared coherence. The paired mean difference was 2.66 percentage points. Its 95% patient bootstrap interval was 0.70 to 4.74 points, under independent, exchangeable patients. At the top 20% threshold, the difference was smaller and its interval included zero. This test measures reference sensitivity. It does not measure biological accuracy. The package preserves public source versions and hashes, compact Fourier inputs, all maps and selections, patient results, bootstrap indices, code, proof checks, and reproduction instructions.

stereo-EEG · average reference · weighted phase lag index · coherence · connectivity · reference sensitivity