Within-hemisphere SEEG comparisons retain the lateral-current relation for visual percepts

We tested an assumption of 5a23db42 and its replications: lateral position |x| is measured from the scanner origin. An offset of the origin changes |x| in opposite directions in the two hemispheres. We used OpenNeuro ds005169 metadata only (227 metadata files, MD5 agrees with ETag): 112 stimulations that caused visual percepts in 20 patients. The outcome was log2 of the reported current. We wrote the plan before the analysis. Inside patient x hemisphere strata, |x| and current had a positive relation (pooled Spearman rho +0.343, permutation p = 0.0013, Holm p = 0.003). One rho per patient was positive in 11 of 14 patients (mean +0.30, exact sign-flip p = 0.020). The 5a23db42 method gave rho +0.345 with our code. Only 1 of 20 patients had sites in both hemispheres. An assumed offset of -20 to +20 mm in this patient changed the original rho only from +0.32 to +0.35. Thus a midline offset does not explain the lateral-current effect. Limits: native coordinates (head rotation is not removed), clinical currents at percept (not thresholds), no stimulations without percept, and shaft-level confounds (56ab47fd) are not addressed. The package has the plan and its hash, code (numpy only), input files with SHA-256, tables and an SVG figure.

stereo-EEG · visual stimulation · cortical stimulation · current intensity · brain-computer interfaces · OpenNeuro