Publication Details
Overview
 
 
 

Unpublished contribution to conference

Abstract 

Background: Accurate electric‐field modeling for transcranial direct‐current stimulation (tDCS)relies on tissue segmentation, and is especially sensitive to how cerebrospinal fluid (CSF) islabelled. We assessed the impact of two CSF segmentation approaches on tDCS simulationresults.Methods: T1‐weighted MRIs from 47 healthy adults (26–69 years) were processed with (i) thedefault CHARM pipeline and (ii) the same pipeline but substituting its CSF mask with one from acommercial pipeline optimized for grey and white matter volumetry. Using SimNIBS 4.1, we rana 1 mA frontal montage (anode F3, cathode F4, 5×5 cm electrodes, 4 mm thick). From corticalgrey matter, the E 99.9 % peak (E99) and the volume experiencing ≥ 75 % of that peak wereextracted. Paired differences failed normality (Shapiro–Wilk), so we applied the Wilcoxonsigned‐rank test and report p-values and effect sizes (r).Results: The second segmentation resulted in larger CSF volume. E99 changed minimally, from0.236 ± 0.058 V/m (CHARM) to 0.224 ± 0.041 V/m (commercial software), median Δ=-0.004V/m (p=0.071, r=0.29). In contrast, focality was more strongly affected: the volume of cortexexperiencing ≥ 75 % of the peak decreased from 10.9 ± 2.3 cm³ to 6.43 ± 2.34 cm³ (medianΔ=-4.16 cm³,≈ 39, p = 3.6 ×10⁻¹⁴) with a large effect size (r = 0.87).Conclusion: As expected, the segmentation with the larger CSF volume, results in a largerspread of the electrical current. This results in lower focality, whereas the strength of theelectrical field remains similar.

Reference