Uncertainty quantification and sensitivity analysis of transcranial electric stimulation for 9-subdomain human head model

Abstract This paper deals with uncertainty quantification of transcranial electric stimulation (TES) of realistic human head model. The head model taken from Visible Human Project consists of 9 subdomains: scalp, skull, CSF, grey matter, white matter, cerebellum, ventricles, jaw and tongue. The deterministic computation of quasi-static induced electric scalar potential features boundary element method (BEM). Conductivities of each subdomain are modelled as uniformly distributed random variables and stochastic analysis features a non-intrusive stochastic collocation method (SCM). The input uncertainties impact only the magnitude of the electric scalar potential and not the position of the potential extrema. Skin and brain conductivities play the most important role, while CSF conductivity has negligible impact on the output potential variance. The significance of the skull conductivity is not high for the chosen input parameter setup. In the previous work authors considered 3-compartment head model which consisted of scalp, skull and brain compartments. The presented model is a step forward in SCM+BEM TES analysis, primarily in terms of model complexity. Comparing the results of the two analyses it can be concluded that the uncertainty in the added tissues’ conductivities do not impact the variation of the output electric potential.