Abstract
Image reconstruction in Electrical Impedance Tomography (EIT) is a nonlinear inverse problem and typically ill-conditioned. A direct consequence of the ill-posedness is high sensitivity errors in measurements. In addition several assumptions made to reduce computational complexity are, in fact, rough approximation. This factors contribute to a limited spatial resolution and result in low accuracy in EIT images. Accordingly to improve electrical impedance images it is necessary to evaluate collected data with a new approach. This paper presents a reconstruction algorithm based on a neural network technique which calculates conductivity changes directly from finite-element simulations of the forward problem. The advantages of this method are it's speed of image reconstruction, it's conceptual simplicity, and ease of implementation.