Priors from Lung Models in EIT Image Reconstruction

Abstract

Discrete Cosine Transfer (DCT) based EIT reconstruction has improved EIT image reconstruction by the addition of prior patient information. In this paper this is furthered by introducing prior patient data into the regularisation term. Three different regularisation terms were used, Tikhonov and two regularisation terms containing DCT coefficients (FDRT and DDRT). The regularisation terms were used for reconstruction of two different lung patterns. The results showed that the DDRT performed better than the FDRT for both lung patterns and both out performed Tikhonov for the gradient lung pattern, showing the potential of introducing prior data into the regularisation term.