Z coordinates show 0.5 mm quantization and larger error in pediatric SEEG contact centres

We assessed z-coordinate quantization and its effect on a second-difference localization-error estimate in 7,267 stereoelectroencephalography (SEEG) contact centres from 60 pediatric patients in OpenNeuro ds005083 v1.0.0. In a prespecified test, 23.2% of z values were within 0.005 mm of a 0.5 mm grid, compared with 1.8% of x and 2.1% of y values; z had a higher grid fraction in 55 of 60 subjects (Holm-adjusted p = 1.5e-5). Across 570 shafts, median per-axis error SD was 0.139 mm in x, 0.116 mm in y, and 0.157 mm in z; z variance was 1.51 times the mean x/y variance and was larger in 44 of 60 subjects (Holm-adjusted p = 1.5e-5). Axial/lateral SD ratios increased with shaft alignment to z (median subject-level Spearman rho +0.11; Holm-adjusted p = 0.010), from 1.09 for |u_z| < 0.3 to 2.13 for |u_z| > 0.6. These results indicate that the axial excess is mostly associated with z-coordinate quantization; the overall 3D error SD remains similar to the earlier estimate (0.240 vs 0.246 mm). A CT-slice localization cause is an assumption because slice spacing and localization software are not reported. The tilt association is correlational, and steep and flat shafts may differ in other ways. For these files, contact distances have extra z error; x and y error, about 0.13 mm per axis, is a better estimate of localization noise.

stereoelectroencephalography · SEEG · electrode localization · coordinate quantization · localization error · intracranial electrodes · neuroimaging · measurement error