Efficient non-linear 3D electrical tomography reconstruction

Abstract

Non-linear electrical tomography imaging can be performed efficiently if certain optimisations are applied to the computational reconstruction process. We present a 3D non-linear reconstruction algorithm based on a regularized conjugate gradient solver and discuss the optimisations which we incorporated to allow for an efficient and accurate reconstruction. In particular, the application of image smoothness constraints or other regularization techniques and auto-adaptive mesh refinement are highly relevant. We demonstrate the results of applying this algorithm to the reconstruction of a simulated material distribution in a cubic volume.