Improving EIT image reconstruction with clustering

In order to improve the image quality for Electrical Impedance Tomography (EIT), the k-means clustering method was adaptively integrated with the standard iterative Gauss-Newton reconstruction algorithm. Within the dual model framework, finite elements were grouped to reduce the numerical complexity in inverse computations. The simulation results indicated that the k-means clustering method did not only preserve the sharp variations over conductivity mediums but also greatly filtered out a simulated additive noise.