Data errors and reconstruction algorithms in electrical impedance tomography.

In electrical impedance tomography the reconstruction problem is a non-linear inverse problem and can only be solved by iterative methods. This paper describes two such algorithms, one based on the regularised Newton's method of Levenburg and Marquardt, and a second modified version of this algorithm which uses optimal current drive patterns. The second algorithm is shown to give superior reconstruction in a simulation study. Some effects of errors in the knowledge of boundary shape and electrode position are also discussed.