An Image Reconstruction Algorithm for Electrical Impedance Tomography Using Adaptive Group Sparsity Constraint

Image quality has long been deemed a key challenge for Electrical Impedance Tomography (EIT). High quality image is of great significance for improving the qualitative and quantitative imaging performance in biomedical or industrial applications. In this paper, a novel image reconstruction algorithm for EIT using adaptive group sparsity constraint is proposed to obtain enhanced image quality. The proposed algorithm takes both the underlying structure characteristics and sparsity prior of the conductivity distribution into account to promote a solution with group sparsity structure and reduce the degree of freedom. Specifically, an adaptive grouping method is incorporated for efficient and dynamic pixel grouping when the conductivity distribution does not have a fixed structure or the prior knowledge of the structure is unavailable. Numerical simulation and phantom experiments are performed to validate the proposed algorithm. The results are compared with those using the Landweber iteration, Total Variation regularization and l1 regularization. Both simulation and experiment results confirm the significantly improved tomographic imaging quality using the proposed algorithm, which demonstrates great potential for multi-phase flow imaging and biological tissue imaging.