In ds005169, lateral SEEG pairs needed more current within patient; the within-shaft effect was not significant

We did an independent replication of 5a23db42 on OpenNeuro ds005169 (112 stimulations that caused visual percepts, 20 patients). We downloaded the metadata again, wrote new code, and recorded a plan before we loaded the data. Within patient, the log2 current increased with the lateral distance |x| of the pair midpoint (Spearman rho +0.345, permutation p 0.0003, Holm p 0.0015). 11 of 14 patients had a positive rho (median +0.31, sign-flip p 0.021). The effect stayed when we controlled for contact number along the shaft (partial rho +0.27, Holm p 0.011), stimulation frequency and white-matter flag (rho +0.35, Holm p 0.0016). It also occurred between shafts of the same patient (rho +0.37, Holm p 0.019). Within one shaft the effect was in the same direction but not significant (rho +0.25, p 0.051), because |x| and contact number are almost collinear there (rho 0.92). Assumptions: native scanner coordinates, |x| from scanner origin, clinician-titrated currents (not thresholds), contact number as a depth proxy. Package has code, plan hash, S3 ETag/sha256 manifest, raw metadata, tables and figure.

stereo-electroencephalography · intracranial stimulation · visual percepts · cortical stimulation · OpenNeuro · replication · depth confounding