Strong beta imaginary coherency residuals had higher mean sign agreement than a label benchmark in paired SEEG files from 36 patients

We tested sign agreement for strong beta imaginary coherency residuals in public OpenNeuro ds004100 version 1.1.3. The data were two interictal SEEG files from each of 36 patients with epilepsy. Each file supplied 60 seconds. Beta was 13 to 30 Hz. We removed row and column means within complete blocks between different shafts. A top 5 percent rule with upward rounding selected the largest absolute residuals in one file. We checked their signs in the other file and averaged both directions. Mean agreement was 58.12 percent. The exact channel-label benchmark was 49.90 percent. Mean excess was 8.21 percentage points. Its 95 percent patient bootstrap interval was 4.02 to 11.49 points. Excess was positive in 34 patients. The fixed mean test gave two-sided p = 0.00007 under independent and sign-symmetric patient effects. All 144 raw ranges passed hash checks. A fresh analysis reproduced all eight result files byte for byte. The package preserves score arrays, selected links, source pins, raw fixtures, proofs, code, figures, and repeat instructions. This study reuses the same cohort as earlier work. The result shows repeated signal map detail. It does not establish neural causes or clinical accuracy.

stereoelectroencephalography · imaginary coherency · sign agreement · connectivity · repeatability · electrophysiology