Construction of anatomy-based priors with anisotropic characteristics with application to electrical impedance tomography

We propose a method for the regularization of biomedical inverse problems and discuss its application to the electrical impedance tomography (EIT). The approach is deterministic but it has a clear statistical interpretation. The method is based on available anatomical information on the investigated object, that can be only of approximative nature. The method is applied to the estimation of the conductivity distribution of the simulated head and it is shown that the implicit statistical characteristics of the regularization method are feasible in this case.