We consider modelling and imaging in EIT under some a priori inequality constraints on conductivity. Instead of constrained optimisation, we reformulate the model equations with respect to some monotone functions that encompass these constraints. We address the cases of positivity and boundness, posing the inverse problem using regularised nonlinear least squares. The results demonstrate significant spatial resolution improvements.
The scalp and skull conductivities (σsc, σsk respectively) are determined from Electrical Impedance Tomography (EIT) data using the Gauss-Newton method (GN). Our best estimates of σsc and σsk are 0.58 S/m and 0.008 S/m respectively. It is necessary to use the true head geometry.