Stochastic Sensitivity Analysis for Dosimetry of Head Tissues for the Three Compartment Head Model

摘要

This paper presents a stochastic framework for the assessment of stochastic sensitivity of electric parameters in the three-compartment model of the human head. The electric parameters of scalp, skull and brain are modelled as random variables with uniform distribution. The propagation of uncertainties from input parameters to the output of interest, i.e. induced electric field is carried out by using the non-intrusive Lagrange stochastic collocation method. The sparse grid interpolation in the multidimensional random space is used to generate the simulation points thus speeding up the calculation compared to traditional Monte Carlo sampling methods or full tensor stochastic collocation methods. The impact of the conductivity and relative permittivity of all three tissues to the induced electric field in the skull and scalp, respectively, is obtained. The presented approach provides a satisfactory insight into the behaviour of the model output with respect to parameter variations and allows the ranking of the input parameters from the most to the least influential ones, respectively.