Pipelined digital processor for dynamical EIT inverse problem solution

In this paper an algorithm for speed-up of the image reconstruction phase in electrical impedance tomography (EIT) systems is presented. The algorithm uses the Kalman filter approach to obtain a linearized model for a dynamical EIT system. However, originally, the system involves stability problems due to the inversion of a stochastic positive semidefinite matrix, which increases with large state dimensions. To overcome that problem, a CR algorithm based on triangular factorization of system matrix is proposed. Even more, the digital signal processor is proposed for implementation of pipeline sparse matrix algorithm demonstrating the optimization results achieved by efficient modeling of sequential algorithm and specific processor hardware with high processor utilization