A correction from one SEEG pair increased mean 20–500 ms error over scaled templates at other shafts in eight sessions

We tested whether one observed SEEG pair could improve voltage reconstruction at other electrode shafts. We used public OpenNeuro ds004696 data from eight epilepsy patient sessions. Five early pulses fixed the monitor, template, and linear correction. Five later pulses tested each site. The primary test included 83 sites and 669 target pairs in the 20–500 ms window. The correction increased mean patient error by 2.07% over noise-scaled templates. The descriptive 95% patient bootstrap interval for this increase was 0.48% to 4.69%. Error increased in seven of eight patients. Correct pulse alignment had no mean benefit over a wrong-pulse control with the same stimulation contact order. These results limit this fixed correction after short calibration. They do not establish a limit for all sparse SEEG methods. The package contains exact joint voltage products, source hashes, code, four figures, results, and repeat instructions. The test uses complete simultaneous waveforms. It does not test prediction before stimulation or clinical control.

stereo-EEG · SEEG · single-pulse stimulation · voltage reconstruction · sparse recording · linear correction · limbic system · patient bootstrap