Bayesian approximation of continuous boundary data for EITMS20

Several direct computational methods (e.g. Layer Stripping, D-bar Method) for reconstructing electric conductivity distributions from current-voltage boundary measurements assume idealized continuous boundary data, while in reality the data consist of measurements from a fixed set of contact electrodes. The passage from discrete measurements to continuous boundary data contributes to the ill-posedness of the EIT inverse problems. In this talk, we propose a computational method rooted in the Bayesian paradigm to estimate continuous data from measurements.

This presentation is part of Minisymposium “MS20 - Advances in Reconstruction Methods for Electrical Impedance Tomography (3 parts)
organized by: Melody Alsaker (Gonzaga University) , Samuli Siltanen (University of Helsinki) .

Sumanth Reddy Nakkireddy (Case Western Reserve University)
bayesian methods, inverse problems, numerical linear algebra